prometheus/tsdb/db_test.go

3748 lines
115 KiB
Go
Raw Normal View History

2017-04-10 11:59:45 -07:00
// Copyright 2017 The Prometheus Authors
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
2016-12-04 04:16:11 -08:00
package tsdb
import (
"bufio"
"context"
"encoding/binary"
"flag"
"fmt"
"hash/crc32"
"math"
"math/rand"
"os"
"path"
"path/filepath"
"sort"
"strconv"
"strings"
"sync"
"testing"
"time"
"github.com/go-kit/log"
"github.com/oklog/ulid"
2017-05-17 07:43:01 -07:00
"github.com/pkg/errors"
"github.com/prometheus/client_golang/prometheus"
prom_testutil "github.com/prometheus/client_golang/prometheus/testutil"
"github.com/stretchr/testify/require"
"go.uber.org/goleak"
"github.com/prometheus/prometheus/model/labels"
Introduce TSDB changes for appending metadata to the WAL (#10972) * Append metadata to the WAL Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove extra whitespace; Reword some docstrings and comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use RLock() for hasNewMetadata check Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use single byte for metric type in RefMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update proposed WAL format for single-byte type metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implementa MetadataAppender interface for the Agent Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of review comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Amend description of metadata in wal.md Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Correct key used to retrieve metadata from cache When we're setting metadata entries in the scrapeCace, we're using the p.Help(), p.Unit(), p.Type() helpers, which retrieve the series name and use it as the cache key. When checking for cache entries though, we used p.Series() as the key, which included the metric name _with_ its labels. That meant that we were never actually hitting the cache. We're fixing this by utiling the __name__ internal label for correctly getting the cache entries after they've been set by setHelp(), setType() or setUnit(). Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Put feature behind a feature flag Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix AppendMetadata docstring Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reorder WAL format document Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Change error message of AppendMetadata; Fix access of s.meta in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse temporary buffer in Metadata encoder Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Only keep latest metadata for each refID during checkpointing Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test that's referencing decoding metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Avoid creating metadata block if no new metadata are present Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for corrupt metadata block and relevant record type Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix CR comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Extract logic about changing metadata in an anonymous function Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implement new proposed WAL format and amend relevant tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use 'const' for metadata field names Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Apply metadata to head memSeries in Commit, not in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add docstring and rename extracted helper in scrape.go Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for tsdb-related cases Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol1 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol2 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix Windows test by closing WAL reader files Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use switch instead of two if statements in metadata decoding Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments around TestMetadata* tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add code for replaying WAL; test correctness of in-memory data after a replay Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove scrape-loop related code from PR Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify tests by sorting slices before comparison Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test to use separate transactions Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Empty out buffer and record slices after encoding latest metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linting issue Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update calculation for DroppedMetadata metric Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Rename MetadataAppender interface and AppendMetadata method to MetadataUpdater/UpdateMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse buffer when encoding latest metadata for each series Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments; Check all returned error values using two helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify use of helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Satisfy linter Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com>
2022-07-19 01:58:52 -07:00
"github.com/prometheus/prometheus/model/metadata"
"github.com/prometheus/prometheus/storage"
"github.com/prometheus/prometheus/tsdb/chunkenc"
"github.com/prometheus/prometheus/tsdb/chunks"
"github.com/prometheus/prometheus/tsdb/fileutil"
"github.com/prometheus/prometheus/tsdb/index"
"github.com/prometheus/prometheus/tsdb/record"
"github.com/prometheus/prometheus/tsdb/tombstones"
"github.com/prometheus/prometheus/tsdb/tsdbutil"
"github.com/prometheus/prometheus/tsdb/wal"
"github.com/prometheus/prometheus/util/testutil"
)
func TestMain(m *testing.M) {
var isolationEnabled bool
flag.BoolVar(&isolationEnabled, "test.tsdb-isolation", true, "enable isolation")
flag.Parse()
defaultIsolationDisabled = !isolationEnabled
goleak.VerifyTestMain(m, goleak.IgnoreTopFunction("github.com/prometheus/prometheus/tsdb.(*SegmentWAL).cut.func1"), goleak.IgnoreTopFunction("github.com/prometheus/prometheus/tsdb.(*SegmentWAL).cut.func2"))
}
func openTestDB(t testing.TB, opts *Options, rngs []int64) (db *DB) {
tmpdir := t.TempDir()
var err error
if len(rngs) == 0 {
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err = Open(tmpdir, nil, nil, opts, nil)
} else {
opts, rngs = validateOpts(opts, rngs)
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err = open(tmpdir, nil, nil, opts, rngs, nil)
}
require.NoError(t, err)
// Do not Close() the test database by default as it will deadlock on test failures.
return db
}
// query runs a matcher query against the querier and fully expands its data.
func query(t testing.TB, q storage.Querier, matchers ...*labels.Matcher) map[string][]tsdbutil.Sample {
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss := q.Select(false, nil, matchers...)
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
defer func() {
require.NoError(t, q.Close())
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
}()
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
result := map[string][]tsdbutil.Sample{}
for ss.Next() {
series := ss.At()
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
samples := []tsdbutil.Sample{}
it := series.Iterator()
for it.Next() {
t, v := it.At()
samples = append(samples, sample{t: t, v: v})
}
require.NoError(t, it.Err())
if len(samples) == 0 {
continue
}
name := series.Labels().String()
result[name] = samples
}
require.NoError(t, ss.Err())
require.Equal(t, 0, len(ss.Warnings()))
return result
}
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
// queryChunks runs a matcher query against the querier and fully expands its data.
func queryChunks(t testing.TB, q storage.ChunkQuerier, matchers ...*labels.Matcher) map[string][]chunks.Meta {
ss := q.Select(false, nil, matchers...)
defer func() {
require.NoError(t, q.Close())
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
}()
result := map[string][]chunks.Meta{}
for ss.Next() {
series := ss.At()
chks := []chunks.Meta{}
it := series.Iterator()
for it.Next() {
chks = append(chks, it.At())
}
require.NoError(t, it.Err())
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
if len(chks) == 0 {
continue
}
name := series.Labels().String()
result[name] = chks
}
require.NoError(t, ss.Err())
require.Equal(t, 0, len(ss.Warnings()))
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
return result
}
// Ensure that blocks are held in memory in their time order
// and not in ULID order as they are read from the directory.
func TestDB_reloadOrder(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
metas := []BlockMeta{
{MinTime: 90, MaxTime: 100},
{MinTime: 70, MaxTime: 80},
{MinTime: 100, MaxTime: 110},
}
for _, m := range metas {
createBlock(t, db.Dir(), genSeries(1, 1, m.MinTime, m.MaxTime))
}
require.NoError(t, db.reloadBlocks())
blocks := db.Blocks()
require.Equal(t, 3, len(blocks))
require.Equal(t, metas[1].MinTime, blocks[0].Meta().MinTime)
require.Equal(t, metas[1].MaxTime, blocks[0].Meta().MaxTime)
require.Equal(t, metas[0].MinTime, blocks[1].Meta().MinTime)
require.Equal(t, metas[0].MaxTime, blocks[1].Meta().MaxTime)
require.Equal(t, metas[2].MinTime, blocks[2].Meta().MinTime)
require.Equal(t, metas[2].MaxTime, blocks[2].Meta().MaxTime)
}
func TestDataAvailableOnlyAfterCommit(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, labels.FromStrings("foo", "bar"), 0, 0)
require.NoError(t, err)
querier, err := db.Querier(context.TODO(), 0, 1)
require.NoError(t, err)
seriesSet := query(t, querier, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
require.Equal(t, map[string][]tsdbutil.Sample{}, seriesSet)
err = app.Commit()
require.NoError(t, err)
querier, err = db.Querier(context.TODO(), 0, 1)
require.NoError(t, err)
defer querier.Close()
seriesSet = query(t, querier, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
require.Equal(t, map[string][]tsdbutil.Sample{`{foo="bar"}`: {sample{t: 0, v: 0}}}, seriesSet)
}
// TestNoPanicAfterWALCorruption ensures that querying the db after a WAL corruption doesn't cause a panic.
// https://github.com/prometheus/prometheus/issues/7548
func TestNoPanicAfterWALCorruption(t *testing.T) {
db := openTestDB(t, &Options{WALSegmentSize: 32 * 1024}, nil)
// Append until the first mmaped head chunk.
// This is to ensure that all samples can be read from the mmaped chunks when the WAL is corrupted.
var expSamples []tsdbutil.Sample
var maxt int64
ctx := context.Background()
{
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
2022-01-10 05:36:45 -08:00
// Appending 121 samples because on the 121st a new chunk will be created.
for i := 0; i < 121; i++ {
app := db.Appender(ctx)
_, err := app.Append(0, labels.FromStrings("foo", "bar"), maxt, 0)
expSamples = append(expSamples, sample{t: maxt, v: 0})
require.NoError(t, err)
require.NoError(t, app.Commit())
maxt++
}
require.NoError(t, db.Close())
}
// Corrupt the WAL after the first sample of the series so that it has at least one sample and
// it is not garbage collected.
// The repair deletes all WAL records after the corrupted record and these are read from the mmaped chunk.
{
walFiles, err := os.ReadDir(path.Join(db.Dir(), "wal"))
require.NoError(t, err)
f, err := os.OpenFile(path.Join(db.Dir(), "wal", walFiles[0].Name()), os.O_RDWR, 0o666)
require.NoError(t, err)
r := wal.NewReader(bufio.NewReader(f))
require.True(t, r.Next(), "reading the series record")
require.True(t, r.Next(), "reading the first sample record")
// Write an invalid record header to corrupt everything after the first wal sample.
_, err = f.WriteAt([]byte{99}, r.Offset())
require.NoError(t, err)
f.Close()
}
// Query the data.
{
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(db.Dir(), nil, nil, nil, nil)
require.NoError(t, err)
defer func() {
require.NoError(t, db.Close())
}()
require.Equal(t, 1.0, prom_testutil.ToFloat64(db.head.metrics.walCorruptionsTotal), "WAL corruption count mismatch")
querier, err := db.Querier(context.TODO(), 0, maxt)
require.NoError(t, err)
seriesSet := query(t, querier, labels.MustNewMatcher(labels.MatchEqual, "", ""))
// The last sample should be missing as it was after the WAL segment corruption.
require.Equal(t, map[string][]tsdbutil.Sample{`{foo="bar"}`: expSamples[0 : len(expSamples)-1]}, seriesSet)
}
}
func TestDataNotAvailableAfterRollback(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
app := db.Appender(context.Background())
_, err := app.Append(0, labels.FromStrings("foo", "bar"), 0, 0)
require.NoError(t, err)
err = app.Rollback()
require.NoError(t, err)
querier, err := db.Querier(context.TODO(), 0, 1)
require.NoError(t, err)
defer querier.Close()
seriesSet := query(t, querier, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
require.Equal(t, map[string][]tsdbutil.Sample{}, seriesSet)
}
func TestDBAppenderAddRef(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app1 := db.Appender(ctx)
ref1, err := app1.Append(0, labels.FromStrings("a", "b"), 123, 0)
require.NoError(t, err)
// Reference should already work before commit.
ref2, err := app1.Append(ref1, labels.EmptyLabels(), 124, 1)
require.NoError(t, err)
require.Equal(t, ref1, ref2)
2017-05-17 07:43:01 -07:00
err = app1.Commit()
require.NoError(t, err)
2017-05-17 07:43:01 -07:00
app2 := db.Appender(ctx)
// first ref should already work in next transaction.
ref3, err := app2.Append(ref1, labels.EmptyLabels(), 125, 0)
require.NoError(t, err)
require.Equal(t, ref1, ref3)
2017-05-17 07:43:01 -07:00
ref4, err := app2.Append(ref1, labels.FromStrings("a", "b"), 133, 1)
require.NoError(t, err)
require.Equal(t, ref1, ref4)
// Reference must be valid to add another sample.
ref5, err := app2.Append(ref2, labels.EmptyLabels(), 143, 2)
require.NoError(t, err)
require.Equal(t, ref1, ref5)
// Missing labels & invalid refs should fail.
_, err = app2.Append(9999999, labels.EmptyLabels(), 1, 1)
require.Equal(t, ErrInvalidSample, errors.Cause(err))
require.NoError(t, app2.Commit())
q, err := db.Querier(context.TODO(), 0, 200)
require.NoError(t, err)
res := query(t, q, labels.MustNewMatcher(labels.MatchEqual, "a", "b"))
require.Equal(t, map[string][]tsdbutil.Sample{
labels.FromStrings("a", "b").String(): {
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
sample{t: 123, v: 0},
sample{t: 124, v: 1},
sample{t: 125, v: 0},
sample{t: 133, v: 1},
sample{t: 143, v: 2},
},
}, res)
}
func TestAppendEmptyLabelsIgnored(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app1 := db.Appender(ctx)
ref1, err := app1.Append(0, labels.FromStrings("a", "b"), 123, 0)
require.NoError(t, err)
// Add with empty label.
ref2, err := app1.Append(0, labels.FromStrings("a", "b", "c", ""), 124, 0)
require.NoError(t, err)
// Should be the same series.
require.Equal(t, ref1, ref2)
err = app1.Commit()
require.NoError(t, err)
}
func TestDeleteSimple(t *testing.T) {
numSamples := int64(10)
cases := []struct {
Intervals tombstones.Intervals
remaint []int64
}{
{
Intervals: tombstones.Intervals{{Mint: 0, Maxt: 3}},
remaint: []int64{4, 5, 6, 7, 8, 9},
},
{
Intervals: tombstones.Intervals{{Mint: 1, Maxt: 3}},
remaint: []int64{0, 4, 5, 6, 7, 8, 9},
},
{
Intervals: tombstones.Intervals{{Mint: 1, Maxt: 3}, {Mint: 4, Maxt: 7}},
remaint: []int64{0, 8, 9},
},
{
Intervals: tombstones.Intervals{{Mint: 1, Maxt: 3}, {Mint: 4, Maxt: 700}},
remaint: []int64{0},
},
{ // This case is to ensure that labels and symbols are deleted.
Intervals: tombstones.Intervals{{Mint: 0, Maxt: 9}},
remaint: []int64{},
},
}
Outer:
for _, c := range cases {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
smpls := make([]float64, numSamples)
for i := int64(0); i < numSamples; i++ {
smpls[i] = rand.Float64()
app.Append(0, labels.FromStrings("a", "b"), i, smpls[i])
}
require.NoError(t, app.Commit())
// TODO(gouthamve): Reset the tombstones somehow.
// Delete the ranges.
for _, r := range c.Intervals {
require.NoError(t, db.Delete(r.Mint, r.Maxt, labels.MustNewMatcher(labels.MatchEqual, "a", "b")))
}
// Compare the result.
q, err := db.Querier(context.TODO(), 0, numSamples)
require.NoError(t, err)
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
res := q.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "a", "b"))
expSamples := make([]tsdbutil.Sample, 0, len(c.remaint))
for _, ts := range c.remaint {
expSamples = append(expSamples, sample{ts, smpls[ts]})
}
expss := newMockSeriesSet([]storage.Series{
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
storage.NewListSeries(labels.FromStrings("a", "b"), expSamples),
})
for {
eok, rok := expss.Next(), res.Next()
require.Equal(t, eok, rok)
if !eok {
require.Equal(t, 0, len(res.Warnings()))
continue Outer
}
sexp := expss.At()
sres := res.At()
require.Equal(t, sexp.Labels(), sres.Labels())
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
smplExp, errExp := storage.ExpandSamples(sexp.Iterator(), nil)
smplRes, errRes := storage.ExpandSamples(sres.Iterator(), nil)
require.Equal(t, errExp, errRes)
require.Equal(t, smplExp, smplRes)
}
}
}
func TestAmendDatapointCausesError(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, labels.FromStrings("a", "b"), 0, 0)
require.NoError(t, err)
require.NoError(t, app.Commit())
app = db.Appender(ctx)
_, err = app.Append(0, labels.FromStrings("a", "b"), 0, 1)
require.Equal(t, storage.ErrDuplicateSampleForTimestamp, err)
require.NoError(t, app.Rollback())
}
func TestDuplicateNaNDatapointNoAmendError(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, labels.FromStrings("a", "b"), 0, math.NaN())
require.NoError(t, err)
require.NoError(t, app.Commit())
app = db.Appender(ctx)
_, err = app.Append(0, labels.FromStrings("a", "b"), 0, math.NaN())
require.NoError(t, err)
}
func TestNonDuplicateNaNDatapointsCausesAmendError(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, labels.FromStrings("a", "b"), 0, math.Float64frombits(0x7ff0000000000001))
require.NoError(t, err)
require.NoError(t, app.Commit())
app = db.Appender(ctx)
_, err = app.Append(0, labels.FromStrings("a", "b"), 0, math.Float64frombits(0x7ff0000000000002))
require.Equal(t, storage.ErrDuplicateSampleForTimestamp, err)
}
func TestEmptyLabelsetCausesError(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, labels.Labels{}, 0, 0)
require.Error(t, err)
require.Equal(t, "empty labelset: invalid sample", err.Error())
}
func TestSkippingInvalidValuesInSameTxn(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
// Append AmendedValue.
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, labels.FromStrings("a", "b"), 0, 1)
require.NoError(t, err)
_, err = app.Append(0, labels.FromStrings("a", "b"), 0, 2)
require.NoError(t, err)
require.NoError(t, app.Commit())
// Make sure the right value is stored.
q, err := db.Querier(context.TODO(), 0, 10)
require.NoError(t, err)
ssMap := query(t, q, labels.MustNewMatcher(labels.MatchEqual, "a", "b"))
require.Equal(t, map[string][]tsdbutil.Sample{
labels.New(labels.Label{Name: "a", Value: "b"}).String(): {sample{0, 1}},
}, ssMap)
// Append Out of Order Value.
app = db.Appender(ctx)
_, err = app.Append(0, labels.FromStrings("a", "b"), 10, 3)
require.NoError(t, err)
_, err = app.Append(0, labels.FromStrings("a", "b"), 7, 5)
require.NoError(t, err)
require.NoError(t, app.Commit())
q, err = db.Querier(context.TODO(), 0, 10)
require.NoError(t, err)
ssMap = query(t, q, labels.MustNewMatcher(labels.MatchEqual, "a", "b"))
require.Equal(t, map[string][]tsdbutil.Sample{
labels.New(labels.Label{Name: "a", Value: "b"}).String(): {sample{0, 1}, sample{10, 3}},
}, ssMap)
}
2017-10-03 05:06:26 -07:00
func TestDB_Snapshot(t *testing.T) {
db := openTestDB(t, nil, nil)
2017-10-03 05:06:26 -07:00
// append data
ctx := context.Background()
app := db.Appender(ctx)
2017-10-03 05:06:26 -07:00
mint := int64(1414141414000)
for i := 0; i < 1000; i++ {
_, err := app.Append(0, labels.FromStrings("foo", "bar"), mint+int64(i), 1.0)
require.NoError(t, err)
2017-10-03 05:06:26 -07:00
}
require.NoError(t, app.Commit())
2017-10-03 05:06:26 -07:00
// create snapshot
snap := t.TempDir()
require.NoError(t, db.Snapshot(snap, true))
require.NoError(t, db.Close())
2017-10-03 05:06:26 -07:00
// reopen DB from snapshot
db, err := Open(snap, nil, nil, nil, nil)
require.NoError(t, err)
defer func() { require.NoError(t, db.Close()) }()
2017-10-03 05:06:26 -07:00
querier, err := db.Querier(context.TODO(), mint, mint+1000)
require.NoError(t, err)
defer func() { require.NoError(t, querier.Close()) }()
2017-10-03 05:06:26 -07:00
// sum values
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
seriesSet := querier.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
2017-10-03 05:06:26 -07:00
sum := 0.0
for seriesSet.Next() {
series := seriesSet.At().Iterator()
for series.Next() {
_, v := series.At()
sum += v
}
require.NoError(t, series.Err())
2017-10-03 05:06:26 -07:00
}
require.NoError(t, seriesSet.Err())
require.Equal(t, 0, len(seriesSet.Warnings()))
require.Equal(t, 1000.0, sum)
2017-10-03 05:06:26 -07:00
}
// TestDB_Snapshot_ChunksOutsideOfCompactedRange ensures that a snapshot removes chunks samples
// that are outside the set block time range.
// See https://github.com/prometheus/prometheus/issues/5105
func TestDB_Snapshot_ChunksOutsideOfCompactedRange(t *testing.T) {
db := openTestDB(t, nil, nil)
ctx := context.Background()
app := db.Appender(ctx)
mint := int64(1414141414000)
for i := 0; i < 1000; i++ {
_, err := app.Append(0, labels.FromStrings("foo", "bar"), mint+int64(i), 1.0)
require.NoError(t, err)
}
require.NoError(t, app.Commit())
snap := t.TempDir()
// Hackingly introduce "race", by having lower max time then maxTime in last chunk.
db.head.maxTime.Sub(10)
require.NoError(t, db.Snapshot(snap, true))
require.NoError(t, db.Close())
// Reopen DB from snapshot.
db, err := Open(snap, nil, nil, nil, nil)
require.NoError(t, err)
defer func() { require.NoError(t, db.Close()) }()
querier, err := db.Querier(context.TODO(), mint, mint+1000)
require.NoError(t, err)
defer func() { require.NoError(t, querier.Close()) }()
// Sum values.
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
seriesSet := querier.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
sum := 0.0
for seriesSet.Next() {
series := seriesSet.At().Iterator()
for series.Next() {
_, v := series.At()
sum += v
}
require.NoError(t, series.Err())
}
require.NoError(t, seriesSet.Err())
require.Equal(t, 0, len(seriesSet.Warnings()))
// Since we snapshotted with MaxTime - 10, so expect 10 less samples.
require.Equal(t, 1000.0-10, sum)
}
func TestDB_SnapshotWithDelete(t *testing.T) {
numSamples := int64(10)
db := openTestDB(t, nil, nil)
defer func() { require.NoError(t, db.Close()) }()
ctx := context.Background()
app := db.Appender(ctx)
smpls := make([]float64, numSamples)
for i := int64(0); i < numSamples; i++ {
smpls[i] = rand.Float64()
app.Append(0, labels.FromStrings("a", "b"), i, smpls[i])
}
require.NoError(t, app.Commit())
cases := []struct {
intervals tombstones.Intervals
remaint []int64
}{
{
intervals: tombstones.Intervals{{Mint: 1, Maxt: 3}, {Mint: 4, Maxt: 7}},
remaint: []int64{0, 8, 9},
},
}
Outer:
for _, c := range cases {
// TODO(gouthamve): Reset the tombstones somehow.
// Delete the ranges.
for _, r := range c.intervals {
require.NoError(t, db.Delete(r.Mint, r.Maxt, labels.MustNewMatcher(labels.MatchEqual, "a", "b")))
}
// create snapshot
snap := t.TempDir()
2018-02-28 03:04:55 -08:00
require.NoError(t, db.Snapshot(snap, true))
// reopen DB from snapshot
newDB, err := Open(snap, nil, nil, nil, nil)
require.NoError(t, err)
defer func() { require.NoError(t, newDB.Close()) }()
// Compare the result.
q, err := newDB.Querier(context.TODO(), 0, numSamples)
require.NoError(t, err)
defer func() { require.NoError(t, q.Close()) }()
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
res := q.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "a", "b"))
expSamples := make([]tsdbutil.Sample, 0, len(c.remaint))
for _, ts := range c.remaint {
expSamples = append(expSamples, sample{ts, smpls[ts]})
}
expss := newMockSeriesSet([]storage.Series{
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
storage.NewListSeries(labels.FromStrings("a", "b"), expSamples),
})
if len(expSamples) == 0 {
require.False(t, res.Next())
continue
}
for {
eok, rok := expss.Next(), res.Next()
require.Equal(t, eok, rok)
if !eok {
require.Equal(t, 0, len(res.Warnings()))
continue Outer
}
sexp := expss.At()
sres := res.At()
require.Equal(t, sexp.Labels(), sres.Labels())
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
smplExp, errExp := storage.ExpandSamples(sexp.Iterator(), nil)
smplRes, errRes := storage.ExpandSamples(sres.Iterator(), nil)
require.Equal(t, errExp, errRes)
require.Equal(t, smplExp, smplRes)
}
}
}
func TestDB_e2e(t *testing.T) {
const (
numDatapoints = 1000
numRanges = 1000
timeInterval = int64(3)
)
// Create 8 series with 1000 data-points of different ranges and run queries.
lbls := [][]labels.Label{
{
{Name: "a", Value: "b"},
{Name: "instance", Value: "localhost:9090"},
{Name: "job", Value: "prometheus"},
},
{
{Name: "a", Value: "b"},
{Name: "instance", Value: "127.0.0.1:9090"},
{Name: "job", Value: "prometheus"},
},
{
{Name: "a", Value: "b"},
{Name: "instance", Value: "127.0.0.1:9090"},
{Name: "job", Value: "prom-k8s"},
},
{
{Name: "a", Value: "b"},
{Name: "instance", Value: "localhost:9090"},
{Name: "job", Value: "prom-k8s"},
},
{
{Name: "a", Value: "c"},
{Name: "instance", Value: "localhost:9090"},
{Name: "job", Value: "prometheus"},
},
{
{Name: "a", Value: "c"},
{Name: "instance", Value: "127.0.0.1:9090"},
{Name: "job", Value: "prometheus"},
},
{
{Name: "a", Value: "c"},
{Name: "instance", Value: "127.0.0.1:9090"},
{Name: "job", Value: "prom-k8s"},
},
{
{Name: "a", Value: "c"},
{Name: "instance", Value: "localhost:9090"},
{Name: "job", Value: "prom-k8s"},
},
}
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
seriesMap := map[string][]tsdbutil.Sample{}
for _, l := range lbls {
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
seriesMap[labels.New(l...).String()] = []tsdbutil.Sample{}
}
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
for _, l := range lbls {
lset := labels.New(l...)
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
series := []tsdbutil.Sample{}
ts := rand.Int63n(300)
for i := 0; i < numDatapoints; i++ {
v := rand.Float64()
series = append(series, sample{ts, v})
_, err := app.Append(0, lset, ts, v)
require.NoError(t, err)
ts += rand.Int63n(timeInterval) + 1
}
seriesMap[lset.String()] = series
}
require.NoError(t, app.Commit())
// Query each selector on 1000 random time-ranges.
queries := []struct {
ms []*labels.Matcher
}{
{
ms: []*labels.Matcher{labels.MustNewMatcher(labels.MatchEqual, "a", "b")},
},
{
ms: []*labels.Matcher{
labels.MustNewMatcher(labels.MatchEqual, "a", "b"),
labels.MustNewMatcher(labels.MatchEqual, "job", "prom-k8s"),
},
},
{
ms: []*labels.Matcher{
labels.MustNewMatcher(labels.MatchEqual, "a", "c"),
labels.MustNewMatcher(labels.MatchEqual, "instance", "localhost:9090"),
labels.MustNewMatcher(labels.MatchEqual, "job", "prometheus"),
},
},
// TODO: Add Regexp Matchers.
}
for _, qry := range queries {
matched := labels.Slice{}
for _, l := range lbls {
s := labels.Selector(qry.ms)
ls := labels.New(l...)
if s.Matches(ls) {
matched = append(matched, ls)
}
}
sort.Sort(matched)
for i := 0; i < numRanges; i++ {
mint := rand.Int63n(300)
maxt := mint + rand.Int63n(timeInterval*int64(numDatapoints))
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
expected := map[string][]tsdbutil.Sample{}
// Build the mockSeriesSet.
for _, m := range matched {
smpls := boundedSamples(seriesMap[m.String()], mint, maxt)
if len(smpls) > 0 {
expected[m.String()] = smpls
}
}
q, err := db.Querier(context.TODO(), mint, maxt)
require.NoError(t, err)
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss := q.Select(false, nil, qry.ms...)
Vertical query merging and compaction (#370) * Vertical series iterator Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Select overlapped blocks first in compactor Plan() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Added vertical compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Code cleanup and comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix tests Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Add benchmark for compaction Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Perform vertical compaction only when blocks are overlapping. Actions for vertical compaction: * Sorting chunk metas * Calling chunks.MergeOverlappingChunks on the chunks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for vertical compaction * BenchmarkNormalCompaction => BenchmarkCompaction * Moved the benchmark from db_test.go to compact_test.go Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Benchmark for query iterator and seek for non overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Vertical query merge only for overlapping blocks Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Simplify logging in Compact(...) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG.md Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Calculate overlapping inside populateBlock Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * MinTime and MaxTime for BlockReader. Using this to find overlapping blocks in populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Sort blocks w.r.t. MinTime in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping in LeveledCompactor.write() instead of returning bool Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Log about overlapping inside LeveledCompactor.populateBlock() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor createBlock to take optional []Series Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * review1 Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com> * Updated CHANGELOG and minor nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * nits Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Updated CHANGELOG Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Refactor iterator and seek benchmarks for Querier. Also has as overlapping blocks. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Additional test case Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * genSeries takes optional labels. Updated BenchmarkQueryIterator and BenchmarkQuerySeek. Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Split genSeries into genSeries and populateSeries Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Check error in benchmark Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Fix review comments Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in> * Warn about overlapping blocks in reload() Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2019-02-14 05:29:41 -08:00
result := map[string][]tsdbutil.Sample{}
for ss.Next() {
x := ss.At()
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
smpls, err := storage.ExpandSamples(x.Iterator(), newSample)
require.NoError(t, err)
if len(smpls) > 0 {
result[x.Labels().String()] = smpls
}
}
require.NoError(t, ss.Err())
require.Equal(t, 0, len(ss.Warnings()))
require.Equal(t, expected, result)
q.Close()
}
}
}
func TestWALFlushedOnDBClose(t *testing.T) {
db := openTestDB(t, nil, nil)
dirDb := db.Dir()
lbls := labels.FromStrings("labelname", "labelvalue")
ctx := context.Background()
app := db.Appender(ctx)
_, err := app.Append(0, lbls, 0, 1)
require.NoError(t, err)
require.NoError(t, app.Commit())
require.NoError(t, db.Close())
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err = Open(dirDb, nil, nil, nil, nil)
require.NoError(t, err)
defer func() { require.NoError(t, db.Close()) }()
q, err := db.Querier(context.TODO(), 0, 1)
require.NoError(t, err)
values, ws, err := q.LabelValues("labelname")
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, []string{"labelvalue"}, values)
}
func TestWALSegmentSizeOptions(t *testing.T) {
tests := map[int]func(dbdir string, segmentSize int){
// Default Wal Size.
0: func(dbDir string, segmentSize int) {
filesAndDir, err := os.ReadDir(filepath.Join(dbDir, "wal"))
require.NoError(t, err)
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
files := []os.FileInfo{}
for _, f := range filesAndDir {
if !f.IsDir() {
fi, err := f.Info()
require.NoError(t, err)
files = append(files, fi)
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
}
}
// All the full segment files (all but the last) should match the segment size option.
for _, f := range files[:len(files)-1] {
require.Equal(t, int64(DefaultOptions().WALSegmentSize), f.Size(), "WAL file size doesn't match WALSegmentSize option, filename: %v", f.Name())
}
lastFile := files[len(files)-1]
require.Greater(t, int64(DefaultOptions().WALSegmentSize), lastFile.Size(), "last WAL file size is not smaller than the WALSegmentSize option, filename: %v", lastFile.Name())
},
// Custom Wal Size.
2 * 32 * 1024: func(dbDir string, segmentSize int) {
filesAndDir, err := os.ReadDir(filepath.Join(dbDir, "wal"))
require.NoError(t, err)
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
files := []os.FileInfo{}
for _, f := range filesAndDir {
if !f.IsDir() {
fi, err := f.Info()
require.NoError(t, err)
files = append(files, fi)
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
}
}
require.Greater(t, len(files), 1, "current WALSegmentSize should result in more than a single WAL file.")
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
// All the full segment files (all but the last) should match the segment size option.
for _, f := range files[:len(files)-1] {
require.Equal(t, int64(segmentSize), f.Size(), "WAL file size doesn't match WALSegmentSize option, filename: %v", f.Name())
}
lastFile := files[len(files)-1]
require.Greater(t, int64(segmentSize), lastFile.Size(), "last WAL file size is not smaller than the WALSegmentSize option, filename: %v", lastFile.Name())
},
// Wal disabled.
-1: func(dbDir string, segmentSize int) {
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
// Check that WAL dir is not there.
_, err := os.Stat(filepath.Join(dbDir, "wal"))
require.Error(t, err)
M-map full chunks of Head from disk (#6679) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. **Prombench results** _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2020-05-06 08:30:00 -07:00
// Check that there is chunks dir.
_, err = os.Stat(mmappedChunksDir(dbDir))
require.NoError(t, err)
},
}
for segmentSize, testFunc := range tests {
t.Run(fmt.Sprintf("WALSegmentSize %d test", segmentSize), func(t *testing.T) {
opts := DefaultOptions()
opts.WALSegmentSize = segmentSize
db := openTestDB(t, opts, nil)
for i := int64(0); i < 155; i++ {
app := db.Appender(context.Background())
ref, err := app.Append(0, labels.FromStrings("wal"+fmt.Sprintf("%d", i), "size"), i, rand.Float64())
require.NoError(t, err)
for j := int64(1); j <= 78; j++ {
_, err := app.Append(ref, labels.EmptyLabels(), i+j, rand.Float64())
require.NoError(t, err)
}
require.NoError(t, app.Commit())
}
dbDir := db.Dir()
require.NoError(t, db.Close())
testFunc(dbDir, int(opts.WALSegmentSize))
})
}
}
// https://github.com/prometheus/prometheus/issues/9846
// https://github.com/prometheus/prometheus/issues/9859
func TestWALReplayRaceOnSamplesLoggedBeforeSeries(t *testing.T) {
const (
numRuns = 1
numSamplesBeforeSeriesCreation = 1000
)
// We test both with few and many samples appended after series creation. If samples are < 120 then there's no
// mmap-ed chunk, otherwise there's at least 1 mmap-ed chunk when replaying the WAL.
for _, numSamplesAfterSeriesCreation := range []int{1, 1000} {
for run := 1; run <= numRuns; run++ {
t.Run(fmt.Sprintf("samples after series creation = %d, run = %d", numSamplesAfterSeriesCreation, run), func(t *testing.T) {
testWALReplayRaceOnSamplesLoggedBeforeSeries(t, numSamplesBeforeSeriesCreation, numSamplesAfterSeriesCreation)
})
}
}
}
func testWALReplayRaceOnSamplesLoggedBeforeSeries(t *testing.T, numSamplesBeforeSeriesCreation, numSamplesAfterSeriesCreation int) {
const numSeries = 1000
db := openTestDB(t, nil, nil)
db.DisableCompactions()
for seriesRef := 1; seriesRef <= numSeries; seriesRef++ {
// Log samples before the series is logged to the WAL.
var enc record.Encoder
var samples []record.RefSample
for ts := 0; ts < numSamplesBeforeSeriesCreation; ts++ {
samples = append(samples, record.RefSample{
Ref: chunks.HeadSeriesRef(uint64(seriesRef)),
T: int64(ts),
V: float64(ts),
})
}
err := db.Head().wal.Log(enc.Samples(samples, nil))
require.NoError(t, err)
// Add samples via appender so that they're logged after the series in the WAL.
app := db.Appender(context.Background())
lbls := labels.FromStrings("series_id", strconv.Itoa(seriesRef))
for ts := numSamplesBeforeSeriesCreation; ts < numSamplesBeforeSeriesCreation+numSamplesAfterSeriesCreation; ts++ {
_, err := app.Append(0, lbls, int64(ts), float64(ts))
require.NoError(t, err)
}
require.NoError(t, app.Commit())
}
require.NoError(t, db.Close())
// Reopen the DB, replaying the WAL.
reopenDB, err := Open(db.Dir(), log.NewLogfmtLogger(os.Stderr), nil, nil, nil)
require.NoError(t, err)
t.Cleanup(func() {
require.NoError(t, reopenDB.Close())
})
// Query back chunks for all series.
q, err := reopenDB.ChunkQuerier(context.Background(), math.MinInt64, math.MaxInt64)
require.NoError(t, err)
set := q.Select(false, nil, labels.MustNewMatcher(labels.MatchRegexp, "series_id", ".+"))
actualSeries := 0
for set.Next() {
actualSeries++
actualChunks := 0
chunksIt := set.At().Iterator()
for chunksIt.Next() {
actualChunks++
}
require.NoError(t, chunksIt.Err())
// We expect 1 chunk every 120 samples after series creation.
require.Equalf(t, (numSamplesAfterSeriesCreation/120)+1, actualChunks, "series: %s", set.At().Labels().String())
}
require.NoError(t, set.Err())
require.Equal(t, numSeries, actualSeries)
}
func TestTombstoneClean(t *testing.T) {
numSamples := int64(10)
db := openTestDB(t, nil, nil)
ctx := context.Background()
app := db.Appender(ctx)
smpls := make([]float64, numSamples)
for i := int64(0); i < numSamples; i++ {
smpls[i] = rand.Float64()
app.Append(0, labels.FromStrings("a", "b"), i, smpls[i])
}
require.NoError(t, app.Commit())
cases := []struct {
intervals tombstones.Intervals
remaint []int64
}{
{
intervals: tombstones.Intervals{{Mint: 1, Maxt: 3}, {Mint: 4, Maxt: 7}},
remaint: []int64{0, 8, 9},
},
}
for _, c := range cases {
// Delete the ranges.
// Create snapshot.
snap := t.TempDir()
require.NoError(t, db.Snapshot(snap, true))
require.NoError(t, db.Close())
// Reopen DB from snapshot.
db, err := Open(snap, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
for _, r := range c.intervals {
require.NoError(t, db.Delete(r.Mint, r.Maxt, labels.MustNewMatcher(labels.MatchEqual, "a", "b")))
}
// All of the setup for THIS line.
require.NoError(t, db.CleanTombstones())
// Compare the result.
q, err := db.Querier(context.TODO(), 0, numSamples)
require.NoError(t, err)
defer q.Close()
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
res := q.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "a", "b"))
expSamples := make([]tsdbutil.Sample, 0, len(c.remaint))
for _, ts := range c.remaint {
expSamples = append(expSamples, sample{ts, smpls[ts]})
}
expss := newMockSeriesSet([]storage.Series{
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
storage.NewListSeries(labels.FromStrings("a", "b"), expSamples),
})
if len(expSamples) == 0 {
require.False(t, res.Next())
continue
}
for {
eok, rok := expss.Next(), res.Next()
require.Equal(t, eok, rok)
if !eok {
break
}
sexp := expss.At()
sres := res.At()
require.Equal(t, sexp.Labels(), sres.Labels())
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
smplExp, errExp := storage.ExpandSamples(sexp.Iterator(), nil)
smplRes, errRes := storage.ExpandSamples(sres.Iterator(), nil)
require.Equal(t, errExp, errRes)
require.Equal(t, smplExp, smplRes)
}
require.Equal(t, 0, len(res.Warnings()))
for _, b := range db.Blocks() {
require.Equal(t, tombstones.NewMemTombstones(), b.tombstones)
}
}
}
// TestTombstoneCleanResultEmptyBlock tests that a TombstoneClean that results in empty blocks (no timeseries)
// will also delete the resultant block.
func TestTombstoneCleanResultEmptyBlock(t *testing.T) {
numSamples := int64(10)
db := openTestDB(t, nil, nil)
ctx := context.Background()
app := db.Appender(ctx)
smpls := make([]float64, numSamples)
for i := int64(0); i < numSamples; i++ {
smpls[i] = rand.Float64()
app.Append(0, labels.FromStrings("a", "b"), i, smpls[i])
}
require.NoError(t, app.Commit())
// Interval should cover the whole block.
intervals := tombstones.Intervals{{Mint: 0, Maxt: numSamples}}
// Create snapshot.
snap := t.TempDir()
require.NoError(t, db.Snapshot(snap, true))
require.NoError(t, db.Close())
// Reopen DB from snapshot.
db, err := Open(snap, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
// Create tombstones by deleting all samples.
for _, r := range intervals {
require.NoError(t, db.Delete(r.Mint, r.Maxt, labels.MustNewMatcher(labels.MatchEqual, "a", "b")))
}
require.NoError(t, db.CleanTombstones())
// After cleaning tombstones that covers the entire block, no blocks should be left behind.
actualBlockDirs, err := blockDirs(db.dir)
require.NoError(t, err)
require.Equal(t, 0, len(actualBlockDirs))
}
// TestTombstoneCleanFail tests that a failing TombstoneClean doesn't leave any blocks behind.
// When TombstoneClean errors the original block that should be rebuilt doesn't get deleted so
// if TombstoneClean leaves any blocks behind these will overlap.
func TestTombstoneCleanFail(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
var oldBlockDirs []string
// Create some blocks pending for compaction.
// totalBlocks should be >=2 so we have enough blocks to trigger compaction failure.
totalBlocks := 2
for i := 0; i < totalBlocks; i++ {
blockDir := createBlock(t, db.Dir(), genSeries(1, 1, int64(i), int64(i)+1))
block, err := OpenBlock(nil, blockDir, nil)
require.NoError(t, err)
// Add some fake tombstones to trigger the compaction.
tomb := tombstones.NewMemTombstones()
tomb.AddInterval(0, tombstones.Interval{Mint: int64(i), Maxt: int64(i) + 1})
block.tombstones = tomb
db.blocks = append(db.blocks, block)
oldBlockDirs = append(oldBlockDirs, blockDir)
}
// Initialize the mockCompactorFailing with a room for a single compaction iteration.
// mockCompactorFailing will fail on the second iteration so we can check if the cleanup works as expected.
db.compactor = &mockCompactorFailing{
t: t,
blocks: db.blocks,
max: totalBlocks + 1,
}
// The compactor should trigger a failure here.
require.Error(t, db.CleanTombstones())
// Now check that the CleanTombstones replaced the old block even after a failure.
actualBlockDirs, err := blockDirs(db.dir)
require.NoError(t, err)
// Only one block should have been replaced by a new block.
require.Equal(t, len(oldBlockDirs), len(actualBlockDirs))
require.Equal(t, len(intersection(oldBlockDirs, actualBlockDirs)), len(actualBlockDirs)-1)
}
// TestTombstoneCleanRetentionLimitsRace tests that a CleanTombstones operation
// and retention limit policies, when triggered at the same time,
// won't race against each other.
func TestTombstoneCleanRetentionLimitsRace(t *testing.T) {
if testing.Short() {
t.Skip("skipping test in short mode.")
}
opts := DefaultOptions()
var wg sync.WaitGroup
// We want to make sure that a race doesn't happen when a normal reload and a CleanTombstones()
// reload try to delete the same block. Without the correct lock placement, it can happen if a
// block is marked for deletion due to retention limits and also has tombstones to be cleaned at
// the same time.
//
// That is something tricky to trigger, so let's try several times just to make sure.
for i := 0; i < 20; i++ {
t.Run(fmt.Sprintf("iteration%d", i), func(t *testing.T) {
db := openTestDB(t, opts, nil)
totalBlocks := 20
dbDir := db.Dir()
// Generate some blocks with old mint (near epoch).
for j := 0; j < totalBlocks; j++ {
blockDir := createBlock(t, dbDir, genSeries(10, 1, int64(j), int64(j)+1))
block, err := OpenBlock(nil, blockDir, nil)
require.NoError(t, err)
// Cover block with tombstones so it can be deleted with CleanTombstones() as well.
tomb := tombstones.NewMemTombstones()
tomb.AddInterval(0, tombstones.Interval{Mint: int64(j), Maxt: int64(j) + 1})
block.tombstones = tomb
db.blocks = append(db.blocks, block)
}
wg.Add(2)
// Run reload and CleanTombstones together, with a small time window randomization
go func() {
defer wg.Done()
time.Sleep(time.Duration(rand.Float64() * 100 * float64(time.Millisecond)))
require.NoError(t, db.reloadBlocks())
}()
go func() {
defer wg.Done()
time.Sleep(time.Duration(rand.Float64() * 100 * float64(time.Millisecond)))
require.NoError(t, db.CleanTombstones())
}()
wg.Wait()
require.NoError(t, db.Close())
})
}
}
func intersection(oldBlocks, actualBlocks []string) (intersection []string) {
hash := make(map[string]bool)
for _, e := range oldBlocks {
hash[e] = true
}
for _, e := range actualBlocks {
// If block present in the hashmap then append intersection list.
if hash[e] {
intersection = append(intersection, e)
}
}
return
}
// mockCompactorFailing creates a new empty block on every write and fails when reached the max allowed total.
type mockCompactorFailing struct {
t *testing.T
blocks []*Block
max int
}
func (*mockCompactorFailing) Plan(dir string) ([]string, error) {
return nil, nil
}
func (c *mockCompactorFailing) Write(dest string, b BlockReader, mint, maxt int64, parent *BlockMeta) (ulid.ULID, error) {
if len(c.blocks) >= c.max {
return ulid.ULID{}, fmt.Errorf("the compactor already did the maximum allowed blocks so it is time to fail")
}
block, err := OpenBlock(nil, createBlock(c.t, dest, genSeries(1, 1, 0, 1)), nil)
require.NoError(c.t, err)
require.NoError(c.t, block.Close()) // Close block as we won't be using anywhere.
c.blocks = append(c.blocks, block)
// Now check that all expected blocks are actually persisted on disk.
// This way we make sure that the we have some blocks that are supposed to be removed.
var expectedBlocks []string
for _, b := range c.blocks {
expectedBlocks = append(expectedBlocks, filepath.Join(dest, b.Meta().ULID.String()))
}
actualBlockDirs, err := blockDirs(dest)
require.NoError(c.t, err)
require.Equal(c.t, expectedBlocks, actualBlockDirs)
return block.Meta().ULID, nil
}
func (*mockCompactorFailing) Compact(string, []string, []*Block) (ulid.ULID, error) {
return ulid.ULID{}, nil
}
func TestTimeRetention(t *testing.T) {
db := openTestDB(t, nil, []int64{1000})
defer func() {
require.NoError(t, db.Close())
}()
blocks := []*BlockMeta{
{MinTime: 500, MaxTime: 900}, // Oldest block
{MinTime: 1000, MaxTime: 1500},
{MinTime: 1500, MaxTime: 2000}, // Newest Block
}
for _, m := range blocks {
createBlock(t, db.Dir(), genSeries(10, 10, m.MinTime, m.MaxTime))
}
require.NoError(t, db.reloadBlocks()) // Reload the db to register the new blocks.
require.Equal(t, len(blocks), len(db.Blocks())) // Ensure all blocks are registered.
2018-02-28 03:04:55 -08:00
db.opts.RetentionDuration = blocks[2].MaxTime - blocks[1].MinTime
require.NoError(t, db.reloadBlocks())
expBlocks := blocks[1:]
actBlocks := db.Blocks()
require.Equal(t, 1, int(prom_testutil.ToFloat64(db.metrics.timeRetentionCount)), "metric retention count mismatch")
require.Equal(t, len(expBlocks), len(actBlocks))
require.Equal(t, expBlocks[0].MaxTime, actBlocks[0].meta.MaxTime)
require.Equal(t, expBlocks[len(expBlocks)-1].MaxTime, actBlocks[len(actBlocks)-1].meta.MaxTime)
}
func TestSizeRetention(t *testing.T) {
db := openTestDB(t, nil, []int64{100})
defer func() {
require.NoError(t, db.Close())
}()
blocks := []*BlockMeta{
{MinTime: 100, MaxTime: 200}, // Oldest block
{MinTime: 200, MaxTime: 300},
{MinTime: 300, MaxTime: 400},
{MinTime: 400, MaxTime: 500},
{MinTime: 500, MaxTime: 600}, // Newest Block
}
2018-02-28 03:04:55 -08:00
for _, m := range blocks {
createBlock(t, db.Dir(), genSeries(100, 10, m.MinTime, m.MaxTime))
}
headBlocks := []*BlockMeta{
{MinTime: 700, MaxTime: 800},
}
// Add some data to the WAL.
headApp := db.Head().Appender(context.Background())
for _, m := range headBlocks {
series := genSeries(100, 10, m.MinTime, m.MaxTime+1)
for _, s := range series {
it := s.Iterator()
for it.Next() {
tim, v := it.At()
_, err := headApp.Append(0, s.Labels(), tim, v)
require.NoError(t, err)
}
require.NoError(t, it.Err())
}
}
require.NoError(t, headApp.Commit())
require.Eventually(t, func() bool {
return db.Head().chunkDiskMapper.IsQueueEmpty()
}, 2*time.Second, 100*time.Millisecond)
// Test that registered size matches the actual disk size.
require.NoError(t, db.reloadBlocks()) // Reload the db to register the new db size.
require.Equal(t, len(blocks), len(db.Blocks())) // Ensure all blocks are registered.
blockSize := int64(prom_testutil.ToFloat64(db.metrics.blocksBytes)) // Use the actual internal metrics.
walSize, err := db.Head().wal.Size()
require.NoError(t, err)
cdmSize, err := db.Head().chunkDiskMapper.Size()
require.NoError(t, err)
require.NotZero(t, cdmSize)
// Expected size should take into account block size + WAL size + Head
// chunks size
expSize := blockSize + walSize + cdmSize
actSize, err := fileutil.DirSize(db.Dir())
require.NoError(t, err)
require.Equal(t, expSize, actSize, "registered size doesn't match actual disk size")
// Create a WAL checkpoint, and compare sizes.
first, last, err := wal.Segments(db.Head().wal.Dir())
require.NoError(t, err)
_, err = wal.Checkpoint(log.NewNopLogger(), db.Head().wal, first, last-1, func(x chunks.HeadSeriesRef) bool { return false }, 0)
require.NoError(t, err)
blockSize = int64(prom_testutil.ToFloat64(db.metrics.blocksBytes)) // Use the actual internal metrics.
walSize, err = db.Head().wal.Size()
require.NoError(t, err)
cdmSize, err = db.Head().chunkDiskMapper.Size()
require.NoError(t, err)
require.NotZero(t, cdmSize)
expSize = blockSize + walSize + cdmSize
actSize, err = fileutil.DirSize(db.Dir())
require.NoError(t, err)
require.Equal(t, expSize, actSize, "registered size doesn't match actual disk size")
// Truncate Chunk Disk Mapper and compare sizes.
require.NoError(t, db.Head().chunkDiskMapper.Truncate(900))
cdmSize, err = db.Head().chunkDiskMapper.Size()
require.NoError(t, err)
require.NotZero(t, cdmSize)
expSize = blockSize + walSize + cdmSize
actSize, err = fileutil.DirSize(db.Dir())
require.NoError(t, err)
require.Equal(t, expSize, actSize, "registered size doesn't match actual disk size")
// Decrease the max bytes limit so that a delete is triggered.
// Check total size, total count and check that the oldest block was deleted.
firstBlockSize := db.Blocks()[0].Size()
sizeLimit := actSize - firstBlockSize
db.opts.MaxBytes = sizeLimit // Set the new db size limit one block smaller that the actual size.
require.NoError(t, db.reloadBlocks()) // Reload the db to register the new db size.
expBlocks := blocks[1:]
actBlocks := db.Blocks()
blockSize = int64(prom_testutil.ToFloat64(db.metrics.blocksBytes))
walSize, err = db.Head().wal.Size()
require.NoError(t, err)
cdmSize, err = db.Head().chunkDiskMapper.Size()
require.NoError(t, err)
require.NotZero(t, cdmSize)
// Expected size should take into account block size + WAL size
expSize = blockSize + walSize + cdmSize
Spelling (#6517) * spelling: alertmanager Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: attributes Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: autocomplete Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: bootstrap Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: caught Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: chunkenc Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: compaction Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: corrupted Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: deletable Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: expected Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: fine-grained Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: initialized Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: iteration Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: javascript Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: multiple Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: number Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: overlapping Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: possible Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: postings Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: procedure Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: programmatic Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: queuing Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: querier Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: repairing Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: received Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: reproducible Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: retention Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: sample Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: segements Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: semantic Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: software [LICENSE] Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: staging Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: timestamp Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: unfortunately Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: uvarint Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: subsequently Signed-off-by: Josh Soref <jsoref@users.noreply.github.com> * spelling: ressamples Signed-off-by: Josh Soref <jsoref@users.noreply.github.com>
2020-01-02 06:54:09 -08:00
actRetentionCount := int(prom_testutil.ToFloat64(db.metrics.sizeRetentionCount))
actSize, err = fileutil.DirSize(db.Dir())
require.NoError(t, err)
require.Equal(t, 1, actRetentionCount, "metric retention count mismatch")
require.Equal(t, actSize, expSize, "metric db size doesn't match actual disk size")
require.LessOrEqual(t, expSize, sizeLimit, "actual size (%v) is expected to be less than or equal to limit (%v)", expSize, sizeLimit)
require.Equal(t, len(blocks)-1, len(actBlocks), "new block count should be decreased from:%v to:%v", len(blocks), len(blocks)-1)
require.Equal(t, expBlocks[0].MaxTime, actBlocks[0].meta.MaxTime, "maxT mismatch of the first block")
require.Equal(t, expBlocks[len(expBlocks)-1].MaxTime, actBlocks[len(actBlocks)-1].meta.MaxTime, "maxT mismatch of the last block")
}
func TestSizeRetentionMetric(t *testing.T) {
cases := []struct {
maxBytes int64
expMaxBytes int64
}{
{maxBytes: 1000, expMaxBytes: 1000},
{maxBytes: 0, expMaxBytes: 0},
{maxBytes: -1000, expMaxBytes: 0},
}
for _, c := range cases {
db := openTestDB(t, &Options{
MaxBytes: c.maxBytes,
}, []int64{100})
defer func() {
require.NoError(t, db.Close())
}()
actMaxBytes := int64(prom_testutil.ToFloat64(db.metrics.maxBytes))
require.Equal(t, actMaxBytes, c.expMaxBytes, "metric retention limit bytes mismatch")
}
}
func TestNotMatcherSelectsLabelsUnsetSeries(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
labelpairs := []labels.Labels{
labels.FromStrings("a", "abcd", "b", "abcde"),
labels.FromStrings("labelname", "labelvalue"),
}
ctx := context.Background()
app := db.Appender(ctx)
for _, lbls := range labelpairs {
_, err := app.Append(0, lbls, 0, 1)
require.NoError(t, err)
}
require.NoError(t, app.Commit())
cases := []struct {
selector labels.Selector
series []labels.Labels
}{{
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchNotEqual, "lname", "lvalue"),
},
series: labelpairs,
}, {
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchEqual, "a", "abcd"),
labels.MustNewMatcher(labels.MatchNotEqual, "b", "abcde"),
},
series: []labels.Labels{},
}, {
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchEqual, "a", "abcd"),
labels.MustNewMatcher(labels.MatchNotEqual, "b", "abc"),
},
series: []labels.Labels{labelpairs[0]},
}, {
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchNotRegexp, "a", "abd.*"),
},
series: labelpairs,
}, {
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchNotRegexp, "a", "abc.*"),
},
series: labelpairs[1:],
}, {
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchNotRegexp, "c", "abd.*"),
},
series: labelpairs,
}, {
selector: labels.Selector{
labels.MustNewMatcher(labels.MatchNotRegexp, "labelname", "labelvalue"),
},
series: labelpairs[:1],
}}
q, err := db.Querier(context.TODO(), 0, 10)
require.NoError(t, err)
defer func() { require.NoError(t, q.Close()) }()
for _, c := range cases {
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss := q.Select(false, nil, c.selector...)
lres, _, ws, err := expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, c.series, lres)
}
}
// expandSeriesSet returns the raw labels in the order they are retrieved from
// the series set and the samples keyed by Labels().String().
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
func expandSeriesSet(ss storage.SeriesSet) ([]labels.Labels, map[string][]sample, storage.Warnings, error) {
resultLabels := []labels.Labels{}
resultSamples := map[string][]sample{}
for ss.Next() {
series := ss.At()
samples := []sample{}
it := series.Iterator()
for it.Next() {
t, v := it.At()
samples = append(samples, sample{t: t, v: v})
}
resultLabels = append(resultLabels, series.Labels())
resultSamples[series.Labels().String()] = samples
}
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
return resultLabels, resultSamples, ss.Warnings(), ss.Err()
}
func TestOverlappingBlocksDetectsAllOverlaps(t *testing.T) {
// Create 10 blocks that does not overlap (0-10, 10-20, ..., 100-110) but in reverse order to ensure our algorithm
// will handle that.
metas := make([]BlockMeta, 11)
for i := 10; i >= 0; i-- {
metas[i] = BlockMeta{MinTime: int64(i * 10), MaxTime: int64((i + 1) * 10)}
}
require.Equal(t, 0, len(OverlappingBlocks(metas)), "we found unexpected overlaps")
// Add overlapping blocks. We've to establish order again since we aren't interested
// in trivial overlaps caused by unorderedness.
add := func(ms ...BlockMeta) []BlockMeta {
repl := append(append([]BlockMeta{}, metas...), ms...)
sort.Slice(repl, func(i, j int) bool {
return repl[i].MinTime < repl[j].MinTime
})
return repl
}
// o1 overlaps with 10-20.
o1 := BlockMeta{MinTime: 15, MaxTime: 17}
require.Equal(t, Overlaps{
{Min: 15, Max: 17}: {metas[1], o1},
}, OverlappingBlocks(add(o1)))
// o2 overlaps with 20-30 and 30-40.
o2 := BlockMeta{MinTime: 21, MaxTime: 31}
require.Equal(t, Overlaps{
{Min: 21, Max: 30}: {metas[2], o2},
{Min: 30, Max: 31}: {o2, metas[3]},
}, OverlappingBlocks(add(o2)))
// o3a and o3b overlaps with 30-40 and each other.
o3a := BlockMeta{MinTime: 33, MaxTime: 39}
o3b := BlockMeta{MinTime: 34, MaxTime: 36}
require.Equal(t, Overlaps{
{Min: 34, Max: 36}: {metas[3], o3a, o3b},
}, OverlappingBlocks(add(o3a, o3b)))
// o4 is 1:1 overlap with 50-60.
o4 := BlockMeta{MinTime: 50, MaxTime: 60}
require.Equal(t, Overlaps{
{Min: 50, Max: 60}: {metas[5], o4},
}, OverlappingBlocks(add(o4)))
// o5 overlaps with 60-70, 70-80 and 80-90.
o5 := BlockMeta{MinTime: 61, MaxTime: 85}
require.Equal(t, Overlaps{
{Min: 61, Max: 70}: {metas[6], o5},
{Min: 70, Max: 80}: {o5, metas[7]},
{Min: 80, Max: 85}: {o5, metas[8]},
}, OverlappingBlocks(add(o5)))
// o6a overlaps with 90-100, 100-110 and o6b, o6b overlaps with 90-100 and o6a.
o6a := BlockMeta{MinTime: 92, MaxTime: 105}
o6b := BlockMeta{MinTime: 94, MaxTime: 99}
require.Equal(t, Overlaps{
{Min: 94, Max: 99}: {metas[9], o6a, o6b},
{Min: 100, Max: 105}: {o6a, metas[10]},
}, OverlappingBlocks(add(o6a, o6b)))
// All together.
require.Equal(t, Overlaps{
{Min: 15, Max: 17}: {metas[1], o1},
{Min: 21, Max: 30}: {metas[2], o2}, {Min: 30, Max: 31}: {o2, metas[3]},
{Min: 34, Max: 36}: {metas[3], o3a, o3b},
{Min: 50, Max: 60}: {metas[5], o4},
{Min: 61, Max: 70}: {metas[6], o5}, {Min: 70, Max: 80}: {o5, metas[7]}, {Min: 80, Max: 85}: {o5, metas[8]},
{Min: 94, Max: 99}: {metas[9], o6a, o6b}, {Min: 100, Max: 105}: {o6a, metas[10]},
}, OverlappingBlocks(add(o1, o2, o3a, o3b, o4, o5, o6a, o6b)))
// Additional case.
var nc1 []BlockMeta
nc1 = append(nc1, BlockMeta{MinTime: 1, MaxTime: 5})
nc1 = append(nc1, BlockMeta{MinTime: 2, MaxTime: 3})
nc1 = append(nc1, BlockMeta{MinTime: 2, MaxTime: 3})
nc1 = append(nc1, BlockMeta{MinTime: 2, MaxTime: 3})
nc1 = append(nc1, BlockMeta{MinTime: 2, MaxTime: 3})
nc1 = append(nc1, BlockMeta{MinTime: 2, MaxTime: 6})
nc1 = append(nc1, BlockMeta{MinTime: 3, MaxTime: 5})
nc1 = append(nc1, BlockMeta{MinTime: 5, MaxTime: 7})
nc1 = append(nc1, BlockMeta{MinTime: 7, MaxTime: 10})
nc1 = append(nc1, BlockMeta{MinTime: 8, MaxTime: 9})
require.Equal(t, Overlaps{
{Min: 2, Max: 3}: {nc1[0], nc1[1], nc1[2], nc1[3], nc1[4], nc1[5]}, // 1-5, 2-3, 2-3, 2-3, 2-3, 2,6
{Min: 3, Max: 5}: {nc1[0], nc1[5], nc1[6]}, // 1-5, 2-6, 3-5
{Min: 5, Max: 6}: {nc1[5], nc1[7]}, // 2-6, 5-7
{Min: 8, Max: 9}: {nc1[8], nc1[9]}, // 7-10, 8-9
}, OverlappingBlocks(nc1))
}
// Regression test for https://github.com/prometheus/tsdb/issues/347
func TestChunkAtBlockBoundary(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
blockRange := db.compactor.(*LeveledCompactor).ranges[0]
label := labels.FromStrings("foo", "bar")
for i := int64(0); i < 3; i++ {
_, err := app.Append(0, label, i*blockRange, 0)
require.NoError(t, err)
_, err = app.Append(0, label, i*blockRange+1000, 0)
require.NoError(t, err)
}
err := app.Commit()
require.NoError(t, err)
err = db.Compact()
require.NoError(t, err)
for _, block := range db.Blocks() {
r, err := block.Index()
require.NoError(t, err)
defer r.Close()
meta := block.Meta()
Reduce memory used by postings offset table. Rather than keeping the offset of each postings list, instead keep the nth offset of the offset of the posting list. As postings list offsets have always been sorted, we can then get to the closest entry before the one we want an iterate forwards. I haven't done much tuning on the 32 number, it was chosen to try not to read through more than a 4k page of data. Switch to a bulk interface for fetching postings. Use it to avoid having to re-read parts of the posting offset table when querying lots of it. For a index with what BenchmarkHeadPostingForMatchers uses RAM for r.postings drops from 3.79MB to 80.19kB or about 48x. Bytes allocated go down by 30%, and suprisingly CPU usage drops by 4-6% for typical queries too. benchmark old ns/op new ns/op delta BenchmarkPostingsForMatchers/Block/n="1"-4 35231 36673 +4.09% BenchmarkPostingsForMatchers/Block/n="1",j="foo"-4 563380 540627 -4.04% BenchmarkPostingsForMatchers/Block/j="foo",n="1"-4 536782 534186 -0.48% BenchmarkPostingsForMatchers/Block/n="1",j!="foo"-4 533990 541550 +1.42% BenchmarkPostingsForMatchers/Block/i=~".*"-4 113374598 117969608 +4.05% BenchmarkPostingsForMatchers/Block/i=~".+"-4 146329884 139651442 -4.56% BenchmarkPostingsForMatchers/Block/i=~""-4 50346510 44961127 -10.70% BenchmarkPostingsForMatchers/Block/i!=""-4 41261550 35356165 -14.31% BenchmarkPostingsForMatchers/Block/n="1",i=~".*",j="foo"-4 112544418 116904010 +3.87% BenchmarkPostingsForMatchers/Block/n="1",i=~".*",i!="2",j="foo"-4 112487086 116864918 +3.89% BenchmarkPostingsForMatchers/Block/n="1",i!=""-4 41094758 35457904 -13.72% BenchmarkPostingsForMatchers/Block/n="1",i!="",j="foo"-4 41906372 36151473 -13.73% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",j="foo"-4 147262414 140424800 -4.64% BenchmarkPostingsForMatchers/Block/n="1",i=~"1.+",j="foo"-4 28615629 27872072 -2.60% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",i!="2",j="foo"-4 147117177 140462403 -4.52% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",i!~"2.*",j="foo"-4 175096826 167902298 -4.11% benchmark old allocs new allocs delta BenchmarkPostingsForMatchers/Block/n="1"-4 4 6 +50.00% BenchmarkPostingsForMatchers/Block/n="1",j="foo"-4 7 11 +57.14% BenchmarkPostingsForMatchers/Block/j="foo",n="1"-4 7 11 +57.14% BenchmarkPostingsForMatchers/Block/n="1",j!="foo"-4 15 17 +13.33% BenchmarkPostingsForMatchers/Block/i=~".*"-4 100010 100012 +0.00% BenchmarkPostingsForMatchers/Block/i=~".+"-4 200069 200040 -0.01% BenchmarkPostingsForMatchers/Block/i=~""-4 200072 200045 -0.01% BenchmarkPostingsForMatchers/Block/i!=""-4 200070 200041 -0.01% BenchmarkPostingsForMatchers/Block/n="1",i=~".*",j="foo"-4 100013 100017 +0.00% BenchmarkPostingsForMatchers/Block/n="1",i=~".*",i!="2",j="foo"-4 100017 100023 +0.01% BenchmarkPostingsForMatchers/Block/n="1",i!=""-4 200073 200046 -0.01% BenchmarkPostingsForMatchers/Block/n="1",i!="",j="foo"-4 200075 200050 -0.01% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",j="foo"-4 200074 200049 -0.01% BenchmarkPostingsForMatchers/Block/n="1",i=~"1.+",j="foo"-4 111165 111150 -0.01% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",i!="2",j="foo"-4 200078 200055 -0.01% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",i!~"2.*",j="foo"-4 311282 311238 -0.01% benchmark old bytes new bytes delta BenchmarkPostingsForMatchers/Block/n="1"-4 264 296 +12.12% BenchmarkPostingsForMatchers/Block/n="1",j="foo"-4 360 424 +17.78% BenchmarkPostingsForMatchers/Block/j="foo",n="1"-4 360 424 +17.78% BenchmarkPostingsForMatchers/Block/n="1",j!="foo"-4 520 552 +6.15% BenchmarkPostingsForMatchers/Block/i=~".*"-4 1600461 1600482 +0.00% BenchmarkPostingsForMatchers/Block/i=~".+"-4 24900801 17259077 -30.69% BenchmarkPostingsForMatchers/Block/i=~""-4 24900836 17259151 -30.69% BenchmarkPostingsForMatchers/Block/i!=""-4 24900760 17259048 -30.69% BenchmarkPostingsForMatchers/Block/n="1",i=~".*",j="foo"-4 1600557 1600621 +0.00% BenchmarkPostingsForMatchers/Block/n="1",i=~".*",i!="2",j="foo"-4 1600717 1600813 +0.01% BenchmarkPostingsForMatchers/Block/n="1",i!=""-4 24900856 17259176 -30.69% BenchmarkPostingsForMatchers/Block/n="1",i!="",j="foo"-4 24900952 17259304 -30.69% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",j="foo"-4 24900993 17259333 -30.69% BenchmarkPostingsForMatchers/Block/n="1",i=~"1.+",j="foo"-4 3788311 3142630 -17.04% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",i!="2",j="foo"-4 24901137 17259509 -30.69% BenchmarkPostingsForMatchers/Block/n="1",i=~".+",i!~"2.*",j="foo"-4 28693086 20405680 -28.88% Signed-off-by: Brian Brazil <brian.brazil@robustperception.io>
2019-12-05 10:27:40 -08:00
k, v := index.AllPostingsKey()
p, err := r.Postings(k, v)
require.NoError(t, err)
var (
lset labels.Labels
chks []chunks.Meta
)
chunkCount := 0
for p.Next() {
err = r.Series(p.At(), &lset, &chks)
require.NoError(t, err)
for _, c := range chks {
require.True(t, meta.MinTime <= c.MinTime && c.MaxTime <= meta.MaxTime,
"chunk spans beyond block boundaries: [block.MinTime=%d, block.MaxTime=%d]; [chunk.MinTime=%d, chunk.MaxTime=%d]",
meta.MinTime, meta.MaxTime, c.MinTime, c.MaxTime)
chunkCount++
}
}
require.Equal(t, 1, chunkCount, "expected 1 chunk in block %s, got %d", meta.ULID, chunkCount)
}
}
func TestQuerierWithBoundaryChunks(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
ctx := context.Background()
app := db.Appender(ctx)
blockRange := db.compactor.(*LeveledCompactor).ranges[0]
label := labels.FromStrings("foo", "bar")
for i := int64(0); i < 5; i++ {
_, err := app.Append(0, label, i*blockRange, 0)
require.NoError(t, err)
_, err = app.Append(0, labels.FromStrings("blockID", strconv.FormatInt(i, 10)), i*blockRange, 0)
require.NoError(t, err)
}
err := app.Commit()
require.NoError(t, err)
err = db.Compact()
require.NoError(t, err)
require.GreaterOrEqual(t, len(db.blocks), 3, "invalid test, less than three blocks in DB")
q, err := db.Querier(context.TODO(), blockRange, 2*blockRange)
require.NoError(t, err)
defer q.Close()
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
// The requested interval covers 2 blocks, so the querier's label values for blockID should give us 2 values, one from each block.
b, ws, err := q.LabelValues("blockID")
require.NoError(t, err)
require.Equal(t, storage.Warnings(nil), ws)
require.Equal(t, []string{"1", "2"}, b)
}
// TestInitializeHeadTimestamp ensures that the h.minTime is set properly.
// - no blocks no WAL: set to the time of the first appended sample
// - no blocks with WAL: set to the smallest sample from the WAL
// - with blocks no WAL: set to the last block maxT
// - with blocks with WAL: same as above
func TestInitializeHeadTimestamp(t *testing.T) {
t.Run("clean", func(t *testing.T) {
dir := t.TempDir()
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(dir, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
// Should be set to init values if no WAL or blocks exist so far.
require.Equal(t, int64(math.MaxInt64), db.head.MinTime())
require.Equal(t, int64(math.MinInt64), db.head.MaxTime())
// First added sample initializes the writable range.
ctx := context.Background()
app := db.Appender(ctx)
_, err = app.Append(0, labels.FromStrings("a", "b"), 1000, 1)
require.NoError(t, err)
require.Equal(t, int64(1000), db.head.MinTime())
require.Equal(t, int64(1000), db.head.MaxTime())
})
t.Run("wal-only", func(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(path.Join(dir, "wal"), 0o777))
w, err := wal.New(nil, nil, path.Join(dir, "wal"), false)
require.NoError(t, err)
var enc record.Encoder
err = w.Log(
enc.Series([]record.RefSeries{
{Ref: 123, Labels: labels.FromStrings("a", "1")},
{Ref: 124, Labels: labels.FromStrings("a", "2")},
}, nil),
enc.Samples([]record.RefSample{
{Ref: 123, T: 5000, V: 1},
{Ref: 124, T: 15000, V: 1},
}, nil),
)
require.NoError(t, err)
require.NoError(t, w.Close())
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(dir, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
require.Equal(t, int64(5000), db.head.MinTime())
require.Equal(t, int64(15000), db.head.MaxTime())
})
t.Run("existing-block", func(t *testing.T) {
dir := t.TempDir()
createBlock(t, dir, genSeries(1, 1, 1000, 2000))
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(dir, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
require.Equal(t, int64(2000), db.head.MinTime())
require.Equal(t, int64(2000), db.head.MaxTime())
})
t.Run("existing-block-and-wal", func(t *testing.T) {
dir := t.TempDir()
createBlock(t, dir, genSeries(1, 1, 1000, 6000))
require.NoError(t, os.MkdirAll(path.Join(dir, "wal"), 0o777))
w, err := wal.New(nil, nil, path.Join(dir, "wal"), false)
require.NoError(t, err)
var enc record.Encoder
err = w.Log(
enc.Series([]record.RefSeries{
{Ref: 123, Labels: labels.FromStrings("a", "1")},
{Ref: 124, Labels: labels.FromStrings("a", "2")},
}, nil),
enc.Samples([]record.RefSample{
{Ref: 123, T: 5000, V: 1},
{Ref: 124, T: 15000, V: 1},
}, nil),
)
require.NoError(t, err)
require.NoError(t, w.Close())
r := prometheus.NewRegistry()
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(dir, nil, r, nil, nil)
require.NoError(t, err)
defer db.Close()
require.Equal(t, int64(6000), db.head.MinTime())
require.Equal(t, int64(15000), db.head.MaxTime())
// Check that old series has been GCed.
require.Equal(t, 1.0, prom_testutil.ToFloat64(db.head.metrics.series))
})
}
func TestNoEmptyBlocks(t *testing.T) {
db := openTestDB(t, nil, []int64{100})
ctx := context.Background()
defer func() {
require.NoError(t, db.Close())
}()
db.DisableCompactions()
rangeToTriggerCompaction := db.compactor.(*LeveledCompactor).ranges[0]/2*3 - 1
defaultLabel := labels.FromStrings("foo", "bar")
defaultMatcher := labels.MustNewMatcher(labels.MatchRegexp, "", ".*")
t.Run("Test no blocks after compact with empty head.", func(t *testing.T) {
require.NoError(t, db.Compact())
actBlocks, err := blockDirs(db.Dir())
require.NoError(t, err)
require.Equal(t, len(db.Blocks()), len(actBlocks))
require.Equal(t, 0, len(actBlocks))
require.Equal(t, 0, int(prom_testutil.ToFloat64(db.compactor.(*LeveledCompactor).metrics.ran)), "no compaction should be triggered here")
})
t.Run("Test no blocks after deleting all samples from head.", func(t *testing.T) {
app := db.Appender(ctx)
_, err := app.Append(0, defaultLabel, 1, 0)
require.NoError(t, err)
_, err = app.Append(0, defaultLabel, 2, 0)
require.NoError(t, err)
_, err = app.Append(0, defaultLabel, 3+rangeToTriggerCompaction, 0)
require.NoError(t, err)
require.NoError(t, app.Commit())
require.NoError(t, db.Delete(math.MinInt64, math.MaxInt64, defaultMatcher))
require.NoError(t, db.Compact())
require.Equal(t, 1, int(prom_testutil.ToFloat64(db.compactor.(*LeveledCompactor).metrics.ran)), "compaction should have been triggered here")
actBlocks, err := blockDirs(db.Dir())
require.NoError(t, err)
require.Equal(t, len(db.Blocks()), len(actBlocks))
require.Equal(t, 0, len(actBlocks))
app = db.Appender(ctx)
_, err = app.Append(0, defaultLabel, 1, 0)
require.Equal(t, storage.ErrOutOfBounds, err, "the head should be truncated so no samples in the past should be allowed")
// Adding new blocks.
currentTime := db.Head().MaxTime()
_, err = app.Append(0, defaultLabel, currentTime, 0)
require.NoError(t, err)
_, err = app.Append(0, defaultLabel, currentTime+1, 0)
require.NoError(t, err)
_, err = app.Append(0, defaultLabel, currentTime+rangeToTriggerCompaction, 0)
require.NoError(t, err)
require.NoError(t, app.Commit())
require.NoError(t, db.Compact())
require.Equal(t, 2, int(prom_testutil.ToFloat64(db.compactor.(*LeveledCompactor).metrics.ran)), "compaction should have been triggered here")
actBlocks, err = blockDirs(db.Dir())
require.NoError(t, err)
require.Equal(t, len(db.Blocks()), len(actBlocks))
require.Equal(t, 1, len(actBlocks), "No blocks created when compacting with >0 samples")
})
t.Run(`When no new block is created from head, and there are some blocks on disk
compaction should not run into infinite loop (was seen during development).`, func(t *testing.T) {
oldBlocks := db.Blocks()
app := db.Appender(ctx)
currentTime := db.Head().MaxTime()
_, err := app.Append(0, defaultLabel, currentTime, 0)
require.NoError(t, err)
_, err = app.Append(0, defaultLabel, currentTime+1, 0)
require.NoError(t, err)
_, err = app.Append(0, defaultLabel, currentTime+rangeToTriggerCompaction, 0)
require.NoError(t, err)
require.NoError(t, app.Commit())
require.NoError(t, db.head.Delete(math.MinInt64, math.MaxInt64, defaultMatcher))
require.NoError(t, db.Compact())
require.Equal(t, 3, int(prom_testutil.ToFloat64(db.compactor.(*LeveledCompactor).metrics.ran)), "compaction should have been triggered here")
require.Equal(t, oldBlocks, db.Blocks())
})
t.Run("Test no blocks remaining after deleting all samples from disk.", func(t *testing.T) {
currentTime := db.Head().MaxTime()
blocks := []*BlockMeta{
{MinTime: currentTime, MaxTime: currentTime + db.compactor.(*LeveledCompactor).ranges[0]},
{MinTime: currentTime + 100, MaxTime: currentTime + 100 + db.compactor.(*LeveledCompactor).ranges[0]},
}
for _, m := range blocks {
createBlock(t, db.Dir(), genSeries(2, 2, m.MinTime, m.MaxTime))
}
oldBlocks := db.Blocks()
require.NoError(t, db.reloadBlocks()) // Reload the db to register the new blocks.
require.Equal(t, len(blocks)+len(oldBlocks), len(db.Blocks())) // Ensure all blocks are registered.
require.NoError(t, db.Delete(math.MinInt64, math.MaxInt64, defaultMatcher))
require.NoError(t, db.Compact())
require.Equal(t, 5, int(prom_testutil.ToFloat64(db.compactor.(*LeveledCompactor).metrics.ran)), "compaction should have been triggered here once for each block that have tombstones")
actBlocks, err := blockDirs(db.Dir())
require.NoError(t, err)
require.Equal(t, len(db.Blocks()), len(actBlocks))
require.Equal(t, 1, len(actBlocks), "All samples are deleted. Only the most recent block should remain after compaction.")
})
}
func TestDB_LabelNames(t *testing.T) {
tests := []struct {
// Add 'sampleLabels1' -> Test Head -> Compact -> Test Disk ->
// -> Add 'sampleLabels2' -> Test Head+Disk
sampleLabels1 [][2]string // For checking head and disk separately.
// To test Head+Disk, sampleLabels2 should have
// at least 1 unique label name which is not in sampleLabels1.
sampleLabels2 [][2]string // // For checking head and disk together.
exp1 []string // after adding sampleLabels1.
exp2 []string // after adding sampleLabels1 and sampleLabels2.
}{
{
sampleLabels1: [][2]string{
{"name1", "1"},
{"name3", "3"},
{"name2", "2"},
},
sampleLabels2: [][2]string{
{"name4", "4"},
{"name1", "1"},
},
exp1: []string{"name1", "name2", "name3"},
exp2: []string{"name1", "name2", "name3", "name4"},
},
{
sampleLabels1: [][2]string{
{"name2", "2"},
{"name1", "1"},
{"name2", "2"},
},
sampleLabels2: [][2]string{
{"name6", "6"},
{"name0", "0"},
},
exp1: []string{"name1", "name2"},
exp2: []string{"name0", "name1", "name2", "name6"},
},
}
blockRange := int64(1000)
// Appends samples into the database.
appendSamples := func(db *DB, mint, maxt int64, sampleLabels [][2]string) {
t.Helper()
ctx := context.Background()
app := db.Appender(ctx)
for i := mint; i <= maxt; i++ {
for _, tuple := range sampleLabels {
label := labels.FromStrings(tuple[0], tuple[1])
_, err := app.Append(0, label, i*blockRange, 0)
require.NoError(t, err)
}
}
err := app.Commit()
require.NoError(t, err)
}
for _, tst := range tests {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
appendSamples(db, 0, 4, tst.sampleLabels1)
// Testing head.
headIndexr, err := db.head.Index()
require.NoError(t, err)
labelNames, err := headIndexr.LabelNames()
require.NoError(t, err)
require.Equal(t, tst.exp1, labelNames)
require.NoError(t, headIndexr.Close())
// Testing disk.
err = db.Compact()
require.NoError(t, err)
// All blocks have same label names, hence check them individually.
// No need to aggregate and check.
for _, b := range db.Blocks() {
blockIndexr, err := b.Index()
require.NoError(t, err)
labelNames, err = blockIndexr.LabelNames()
require.NoError(t, err)
require.Equal(t, tst.exp1, labelNames)
require.NoError(t, blockIndexr.Close())
}
// Adding more samples to head with new label names
// so that we can test (head+disk).LabelNames() (the union).
appendSamples(db, 5, 9, tst.sampleLabels2)
// Testing DB (union).
q, err := db.Querier(context.TODO(), math.MinInt64, math.MaxInt64)
require.NoError(t, err)
var ws storage.Warnings
labelNames, ws, err = q.LabelNames()
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.NoError(t, q.Close())
require.Equal(t, tst.exp2, labelNames)
}
}
func TestCorrectNumTombstones(t *testing.T) {
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
blockRange := db.compactor.(*LeveledCompactor).ranges[0]
name, value := "foo", "bar"
defaultLabel := labels.FromStrings(name, value)
defaultMatcher := labels.MustNewMatcher(labels.MatchEqual, name, value)
ctx := context.Background()
app := db.Appender(ctx)
for i := int64(0); i < 3; i++ {
for j := int64(0); j < 15; j++ {
_, err := app.Append(0, defaultLabel, i*blockRange+j, 0)
require.NoError(t, err)
}
}
require.NoError(t, app.Commit())
err := db.Compact()
require.NoError(t, err)
require.Equal(t, 1, len(db.blocks))
require.NoError(t, db.Delete(0, 1, defaultMatcher))
require.Equal(t, uint64(1), db.blocks[0].meta.Stats.NumTombstones)
// {0, 1} and {2, 3} are merged to form 1 tombstone.
require.NoError(t, db.Delete(2, 3, defaultMatcher))
require.Equal(t, uint64(1), db.blocks[0].meta.Stats.NumTombstones)
require.NoError(t, db.Delete(5, 6, defaultMatcher))
require.Equal(t, uint64(2), db.blocks[0].meta.Stats.NumTombstones)
require.NoError(t, db.Delete(9, 11, defaultMatcher))
require.Equal(t, uint64(3), db.blocks[0].meta.Stats.NumTombstones)
}
// TestBlockRanges checks the following use cases:
// - No samples can be added with timestamps lower than the last block maxt.
// - The compactor doesn't create overlapping blocks
//
// even when the last blocks is not within the default boundaries.
// - Lower boundary is based on the smallest sample in the head and
//
// upper boundary is rounded to the configured block range.
//
// This ensures that a snapshot that includes the head and creates a block with a custom time range
// will not overlap with the first block created by the next compaction.
func TestBlockRanges(t *testing.T) {
logger := log.NewLogfmtLogger(log.NewSyncWriter(os.Stderr))
ctx := context.Background()
dir := t.TempDir()
// Test that the compactor doesn't create overlapping blocks
// when a non standard block already exists.
firstBlockMaxT := int64(3)
createBlock(t, dir, genSeries(1, 1, 0, firstBlockMaxT))
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := open(dir, logger, nil, DefaultOptions(), []int64{10000}, nil)
require.NoError(t, err)
rangeToTriggerCompaction := db.compactor.(*LeveledCompactor).ranges[0]/2*3 + 1
app := db.Appender(ctx)
lbl := labels.FromStrings("a", "b")
_, err = app.Append(0, lbl, firstBlockMaxT-1, rand.Float64())
if err == nil {
t.Fatalf("appending a sample with a timestamp covered by a previous block shouldn't be possible")
}
_, err = app.Append(0, lbl, firstBlockMaxT+1, rand.Float64())
require.NoError(t, err)
_, err = app.Append(0, lbl, firstBlockMaxT+2, rand.Float64())
require.NoError(t, err)
secondBlockMaxt := firstBlockMaxT + rangeToTriggerCompaction
_, err = app.Append(0, lbl, secondBlockMaxt, rand.Float64()) // Add samples to trigger a new compaction
require.NoError(t, err)
require.NoError(t, app.Commit())
for x := 0; x < 100; x++ {
if len(db.Blocks()) == 2 {
break
}
time.Sleep(100 * time.Millisecond)
}
require.Equal(t, 2, len(db.Blocks()), "no new block created after the set timeout")
if db.Blocks()[0].Meta().MaxTime > db.Blocks()[1].Meta().MinTime {
t.Fatalf("new block overlaps old:%v,new:%v", db.Blocks()[0].Meta(), db.Blocks()[1].Meta())
}
// Test that wal records are skipped when an existing block covers the same time ranges
// and compaction doesn't create an overlapping block.
app = db.Appender(ctx)
db.DisableCompactions()
_, err = app.Append(0, lbl, secondBlockMaxt+1, rand.Float64())
require.NoError(t, err)
_, err = app.Append(0, lbl, secondBlockMaxt+2, rand.Float64())
require.NoError(t, err)
_, err = app.Append(0, lbl, secondBlockMaxt+3, rand.Float64())
require.NoError(t, err)
_, err = app.Append(0, lbl, secondBlockMaxt+4, rand.Float64())
require.NoError(t, err)
require.NoError(t, app.Commit())
require.NoError(t, db.Close())
thirdBlockMaxt := secondBlockMaxt + 2
createBlock(t, dir, genSeries(1, 1, secondBlockMaxt+1, thirdBlockMaxt))
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err = open(dir, logger, nil, DefaultOptions(), []int64{10000}, nil)
require.NoError(t, err)
defer db.Close()
require.Equal(t, 3, len(db.Blocks()), "db doesn't include expected number of blocks")
require.Equal(t, db.Blocks()[2].Meta().MaxTime, thirdBlockMaxt, "unexpected maxt of the last block")
app = db.Appender(ctx)
_, err = app.Append(0, lbl, thirdBlockMaxt+rangeToTriggerCompaction, rand.Float64()) // Trigger a compaction
require.NoError(t, err)
require.NoError(t, app.Commit())
for x := 0; x < 100; x++ {
if len(db.Blocks()) == 4 {
break
}
time.Sleep(100 * time.Millisecond)
}
require.Equal(t, 4, len(db.Blocks()), "no new block created after the set timeout")
if db.Blocks()[2].Meta().MaxTime > db.Blocks()[3].Meta().MinTime {
t.Fatalf("new block overlaps old:%v,new:%v", db.Blocks()[2].Meta(), db.Blocks()[3].Meta())
}
}
// TestDBReadOnly ensures that opening a DB in readonly mode doesn't modify any files on the disk.
// It also checks that the API calls return equivalent results as a normal db.Open() mode.
func TestDBReadOnly(t *testing.T) {
var (
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
dbDir string
logger = log.NewLogfmtLogger(log.NewSyncWriter(os.Stderr))
expBlocks []*Block
expSeries map[string][]tsdbutil.Sample
expChunks map[string][]chunks.Meta
expDBHash []byte
matchAll = labels.MustNewMatcher(labels.MatchEqual, "", "")
err error
)
// Bootstrap the db.
{
dbDir = t.TempDir()
dbBlocks := []*BlockMeta{
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
// Create three 2-sample blocks.
{MinTime: 10, MaxTime: 12},
{MinTime: 12, MaxTime: 14},
{MinTime: 14, MaxTime: 16},
}
for _, m := range dbBlocks {
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
_ = createBlock(t, dbDir, genSeries(1, 1, m.MinTime, m.MaxTime))
}
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
// Add head to test DBReadOnly WAL reading capabilities.
w, err := wal.New(logger, nil, filepath.Join(dbDir, "wal"), true)
require.NoError(t, err)
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
h := createHead(t, w, genSeries(1, 1, 16, 18), dbDir)
require.NoError(t, h.Close())
}
// Open a normal db to use for a comparison.
{
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
dbWritable, err := Open(dbDir, logger, nil, nil, nil)
require.NoError(t, err)
dbWritable.DisableCompactions()
dbSizeBeforeAppend, err := fileutil.DirSize(dbWritable.Dir())
require.NoError(t, err)
app := dbWritable.Appender(context.Background())
_, err = app.Append(0, labels.FromStrings("foo", "bar"), dbWritable.Head().MaxTime()+1, 0)
require.NoError(t, err)
require.NoError(t, app.Commit())
expBlocks = dbWritable.Blocks()
expDbSize, err := fileutil.DirSize(dbWritable.Dir())
require.NoError(t, err)
require.Greater(t, expDbSize, dbSizeBeforeAppend, "db size didn't increase after an append")
q, err := dbWritable.Querier(context.TODO(), math.MinInt64, math.MaxInt64)
require.NoError(t, err)
expSeries = query(t, q, matchAll)
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
cq, err := dbWritable.ChunkQuerier(context.TODO(), math.MinInt64, math.MaxInt64)
require.NoError(t, err)
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
expChunks = queryChunks(t, cq, matchAll)
require.NoError(t, dbWritable.Close()) // Close here to allow getting the dir hash for windows.
expDBHash = testutil.DirHash(t, dbWritable.Dir())
}
// Open a read only db and ensure that the API returns the same result as the normal DB.
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
dbReadOnly, err := OpenDBReadOnly(dbDir, logger)
require.NoError(t, err)
defer func() { require.NoError(t, dbReadOnly.Close()) }()
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
t.Run("blocks", func(t *testing.T) {
blocks, err := dbReadOnly.Blocks()
require.NoError(t, err)
require.Equal(t, len(expBlocks), len(blocks))
for i, expBlock := range expBlocks {
require.Equal(t, expBlock.Meta(), blocks[i].Meta(), "block meta mismatch")
}
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
})
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
t.Run("querier", func(t *testing.T) {
// Open a read only db and ensure that the API returns the same result as the normal DB.
q, err := dbReadOnly.Querier(context.TODO(), math.MinInt64, math.MaxInt64)
require.NoError(t, err)
readOnlySeries := query(t, q, matchAll)
readOnlyDBHash := testutil.DirHash(t, dbDir)
require.Equal(t, len(expSeries), len(readOnlySeries), "total series mismatch")
require.Equal(t, expSeries, readOnlySeries, "series mismatch")
require.Equal(t, expDBHash, readOnlyDBHash, "after all read operations the db hash should remain the same")
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
})
t.Run("chunk querier", func(t *testing.T) {
cq, err := dbReadOnly.ChunkQuerier(context.TODO(), math.MinInt64, math.MaxInt64)
require.NoError(t, err)
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
readOnlySeries := queryChunks(t, cq, matchAll)
readOnlyDBHash := testutil.DirHash(t, dbDir)
require.Equal(t, len(expChunks), len(readOnlySeries), "total series mismatch")
require.Equal(t, expChunks, readOnlySeries, "series chunks mismatch")
require.Equal(t, expDBHash, readOnlyDBHash, "after all read operations the db hash should remain the same")
tsdb: Added ChunkQueryable implementations to db; unified MergeSeriesSets and vertical to single struct. (#7069) * tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating. Chained to https://github.com/prometheus/prometheus/pull/7059 * NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it. * Added single SeriesEntry / ChunkEntry for all series implementations. * Unified all vertical, and non vertical for compact and querying to single merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before) * Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples. * Refactored endpoint tests and querier tests to include subtests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed comments from Brian and Beorn. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed snapshot test and added chunk iterator support for DBReadOnly. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed race when iterating over Ats first. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed populate block tests. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed endpoints test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed test. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added test & fixed case of head open chunk. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed DBReadOnly tests and bug producing 1 sample chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added cases for partial block overlap for multiple full chunks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Added extra tests for chunk meta after compaction. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fixed small vertical merge bug and added more tests for that. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-07-31 08:03:02 -07:00
})
}
// TestDBReadOnlyClosing ensures that after closing the db
// all api methods return an ErrClosed.
func TestDBReadOnlyClosing(t *testing.T) {
dbDir := t.TempDir()
db, err := OpenDBReadOnly(dbDir, log.NewLogfmtLogger(log.NewSyncWriter(os.Stderr)))
require.NoError(t, err)
require.NoError(t, db.Close())
require.Equal(t, db.Close(), ErrClosed)
_, err = db.Blocks()
require.Equal(t, err, ErrClosed)
_, err = db.Querier(context.TODO(), 0, 1)
require.Equal(t, err, ErrClosed)
}
func TestDBReadOnly_FlushWAL(t *testing.T) {
var (
dbDir string
logger = log.NewLogfmtLogger(log.NewSyncWriter(os.Stderr))
err error
maxt int
ctx = context.Background()
)
// Bootstrap the db.
{
dbDir = t.TempDir()
// Append data to the WAL.
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(dbDir, logger, nil, nil, nil)
require.NoError(t, err)
db.DisableCompactions()
app := db.Appender(ctx)
maxt = 1000
for i := 0; i < maxt; i++ {
_, err := app.Append(0, labels.FromStrings(defaultLabelName, "flush"), int64(i), 1.0)
require.NoError(t, err)
}
require.NoError(t, app.Commit())
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
2022-01-10 05:36:45 -08:00
require.NoError(t, db.Close())
}
// Flush WAL.
db, err := OpenDBReadOnly(dbDir, logger)
require.NoError(t, err)
flush := t.TempDir()
require.NoError(t, db.FlushWAL(flush))
require.NoError(t, db.Close())
// Reopen the DB from the flushed WAL block.
db, err = OpenDBReadOnly(flush, logger)
require.NoError(t, err)
defer func() { require.NoError(t, db.Close()) }()
blocks, err := db.Blocks()
require.NoError(t, err)
require.Equal(t, len(blocks), 1)
querier, err := db.Querier(context.TODO(), 0, int64(maxt)-1)
require.NoError(t, err)
defer func() { require.NoError(t, querier.Close()) }()
// Sum the values.
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
seriesSet := querier.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, defaultLabelName, "flush"))
sum := 0.0
for seriesSet.Next() {
series := seriesSet.At().Iterator()
for series.Next() {
_, v := series.At()
sum += v
}
require.NoError(t, series.Err())
}
require.NoError(t, seriesSet.Err())
require.Equal(t, 0, len(seriesSet.Warnings()))
require.Equal(t, 1000.0, sum)
}
func TestDBCannotSeePartialCommits(t *testing.T) {
if defaultIsolationDisabled {
t.Skip("skipping test since tsdb isolation is disabled")
}
tmpdir := t.TempDir()
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(tmpdir, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
stop := make(chan struct{})
firstInsert := make(chan struct{})
ctx := context.Background()
// Insert data in batches.
go func() {
iter := 0
for {
app := db.Appender(ctx)
for j := 0; j < 100; j++ {
_, err := app.Append(0, labels.FromStrings("foo", "bar", "a", strconv.Itoa(j)), int64(iter), float64(iter))
require.NoError(t, err)
}
err = app.Commit()
require.NoError(t, err)
if iter == 0 {
close(firstInsert)
}
iter++
select {
case <-stop:
return
default:
}
}
}()
<-firstInsert
// This is a race condition, so do a few tests to tickle it.
// Usually most will fail.
inconsistencies := 0
for i := 0; i < 10; i++ {
func() {
querier, err := db.Querier(context.Background(), 0, 1000000)
require.NoError(t, err)
defer querier.Close()
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss := querier.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
_, seriesSet, ws, err := expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
values := map[float64]struct{}{}
for _, series := range seriesSet {
values[series[len(series)-1].v] = struct{}{}
}
if len(values) != 1 {
inconsistencies++
}
}()
}
stop <- struct{}{}
require.Equal(t, 0, inconsistencies, "Some queries saw inconsistent results.")
}
func TestDBQueryDoesntSeeAppendsAfterCreation(t *testing.T) {
if defaultIsolationDisabled {
t.Skip("skipping test since tsdb isolation is disabled")
}
tmpdir := t.TempDir()
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(tmpdir, nil, nil, nil, nil)
require.NoError(t, err)
defer db.Close()
querierBeforeAdd, err := db.Querier(context.Background(), 0, 1000000)
require.NoError(t, err)
defer querierBeforeAdd.Close()
ctx := context.Background()
app := db.Appender(ctx)
_, err = app.Append(0, labels.FromStrings("foo", "bar"), 0, 0)
require.NoError(t, err)
querierAfterAddButBeforeCommit, err := db.Querier(context.Background(), 0, 1000000)
require.NoError(t, err)
defer querierAfterAddButBeforeCommit.Close()
// None of the queriers should return anything after the Add but before the commit.
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss := querierBeforeAdd.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
_, seriesSet, ws, err := expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, map[string][]sample{}, seriesSet)
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss = querierAfterAddButBeforeCommit.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
_, seriesSet, ws, err = expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, map[string][]sample{}, seriesSet)
// This commit is after the queriers are created, so should not be returned.
err = app.Commit()
require.NoError(t, err)
// Nothing returned for querier created before the Add.
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss = querierBeforeAdd.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
_, seriesSet, ws, err = expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, map[string][]sample{}, seriesSet)
// Series exists but has no samples for querier created after Add.
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss = querierAfterAddButBeforeCommit.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
_, seriesSet, ws, err = expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, map[string][]sample{`{foo="bar"}`: {}}, seriesSet)
querierAfterCommit, err := db.Querier(context.Background(), 0, 1000000)
require.NoError(t, err)
defer querierAfterCommit.Close()
// Samples are returned for querier created after Commit.
*: Consistent Error/Warning handling for SeriesSet iterator: Allowing Async Select (#7251) * Add errors and Warnings to SeriesSet Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Change Querier interface and refactor accordingly Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor promql/engine to propagate warnings at eval stage Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Make sure all the series from all Selects are pre-advanced Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Separate merge series sets Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Clean Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactor merge querier failure handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Refactored and simplified fanout with improvements from incoming chunk iterator PRs. * Secondary logic is hidden, instead of weird failed series set logic we had. * Fanout is well commented * Fanout closing record all errors * MergeQuerier improved API (clearer) * deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Fix formatting Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix CI issues Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Added final tests for error handling. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Addressed Brian's comments. * Moved hints in populate to be allocated only when needed. * Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic. * Select after first Next is done will panic. NOTE: in lazySeriesSet in theory we could just panic, I think however we can totally just return error, it will panic in expand anyway. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Utilize errWithWarnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix recently introduced expansion issue Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add tests for secondary querier error handling Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Implement lazy merge Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Add name to test cases Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Reorganize Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Address review comments Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Remove redundant warnings Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> * Fix rebase mistake Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2020-06-09 09:57:31 -07:00
ss = querierAfterCommit.Select(false, nil, labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"))
_, seriesSet, ws, err = expandSeriesSet(ss)
require.NoError(t, err)
require.Equal(t, 0, len(ws))
require.Equal(t, map[string][]sample{`{foo="bar"}`: {{t: 0, v: 0}}}, seriesSet)
}
// TestChunkWriter_ReadAfterWrite ensures that chunk segment are cut at the set segment size and
// that the resulted segments includes the expected chunks data.
func TestChunkWriter_ReadAfterWrite(t *testing.T) {
chk1 := tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 1}})
chk2 := tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 2}})
chk3 := tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 3}})
chk4 := tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 4}})
chk5 := tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 5}})
chunkSize := len(chk1.Chunk.Bytes()) + chunks.MaxChunkLengthFieldSize + chunks.ChunkEncodingSize + crc32.Size
tests := []struct {
chks [][]chunks.Meta
segmentSize,
expSegmentsCount int
expSegmentSizes []int
}{
// 0:Last chunk ends at the segment boundary so
// all chunks should fit in a single segment.
{
chks: [][]chunks.Meta{
{
chk1,
chk2,
chk3,
},
},
segmentSize: 3 * chunkSize,
expSegmentSizes: []int{3 * chunkSize},
expSegmentsCount: 1,
},
// 1:Two chunks can fit in a single segment so the last one should result in a new segment.
{
chks: [][]chunks.Meta{
{
chk1,
chk2,
chk3,
chk4,
chk5,
},
},
segmentSize: 2 * chunkSize,
expSegmentSizes: []int{2 * chunkSize, 2 * chunkSize, chunkSize},
expSegmentsCount: 3,
},
// 2:When the segment size is smaller than the size of 2 chunks
// the last segment should still create a new segment.
{
chks: [][]chunks.Meta{
{
chk1,
chk2,
chk3,
},
},
segmentSize: 2*chunkSize - 1,
expSegmentSizes: []int{chunkSize, chunkSize, chunkSize},
expSegmentsCount: 3,
},
// 3:When the segment is smaller than a single chunk
// it should still be written by ignoring the max segment size.
{
chks: [][]chunks.Meta{
{
chk1,
},
},
segmentSize: chunkSize - 1,
expSegmentSizes: []int{chunkSize},
expSegmentsCount: 1,
},
// 4:All chunks are bigger than the max segment size, but
// these should still be written even when this will result in bigger segment than the set size.
// Each segment will hold a single chunk.
{
chks: [][]chunks.Meta{
{
chk1,
chk2,
chk3,
},
},
segmentSize: 1,
expSegmentSizes: []int{chunkSize, chunkSize, chunkSize},
expSegmentsCount: 3,
},
// 5:Adding multiple batches of chunks.
{
chks: [][]chunks.Meta{
{
chk1,
chk2,
chk3,
},
{
chk4,
chk5,
},
},
segmentSize: 3 * chunkSize,
expSegmentSizes: []int{3 * chunkSize, 2 * chunkSize},
expSegmentsCount: 2,
},
// 6:Adding multiple batches of chunks.
{
chks: [][]chunks.Meta{
{
chk1,
},
{
chk2,
chk3,
},
{
chk4,
},
},
segmentSize: 2 * chunkSize,
expSegmentSizes: []int{2 * chunkSize, 2 * chunkSize},
expSegmentsCount: 2,
},
}
for i, test := range tests {
t.Run(strconv.Itoa(i), func(t *testing.T) {
tempDir := t.TempDir()
chunkw, err := chunks.NewWriterWithSegSize(tempDir, chunks.SegmentHeaderSize+int64(test.segmentSize))
require.NoError(t, err)
for _, chks := range test.chks {
require.NoError(t, chunkw.WriteChunks(chks...))
}
require.NoError(t, chunkw.Close())
files, err := os.ReadDir(tempDir)
require.NoError(t, err)
require.Equal(t, test.expSegmentsCount, len(files), "expected segments count mismatch")
// Verify that all data is written to the segments.
sizeExp := 0
sizeAct := 0
for _, chks := range test.chks {
for _, chk := range chks {
l := make([]byte, binary.MaxVarintLen32)
sizeExp += binary.PutUvarint(l, uint64(len(chk.Chunk.Bytes()))) // The length field.
sizeExp += chunks.ChunkEncodingSize
sizeExp += len(chk.Chunk.Bytes()) // The data itself.
sizeExp += crc32.Size // The 4 bytes of crc32
}
}
sizeExp += test.expSegmentsCount * chunks.SegmentHeaderSize // The segment header bytes.
for i, f := range files {
fi, err := f.Info()
require.NoError(t, err)
size := int(fi.Size())
// Verify that the segment is the same or smaller than the expected size.
require.GreaterOrEqual(t, chunks.SegmentHeaderSize+test.expSegmentSizes[i], size, "Segment:%v should NOT be bigger than:%v actual:%v", i, chunks.SegmentHeaderSize+test.expSegmentSizes[i], size)
sizeAct += size
}
require.Equal(t, sizeExp, sizeAct)
// Check the content of the chunks.
r, err := chunks.NewDirReader(tempDir, nil)
require.NoError(t, err)
defer func() { require.NoError(t, r.Close()) }()
for _, chks := range test.chks {
for _, chkExp := range chks {
chkAct, err := r.Chunk(chkExp.Ref)
require.NoError(t, err)
require.Equal(t, chkExp.Chunk.Bytes(), chkAct.Bytes())
}
}
})
}
}
func TestRangeForTimestamp(t *testing.T) {
type args struct {
t int64
width int64
}
tests := []struct {
args args
expected int64
}{
{args{0, 5}, 5},
{args{1, 5}, 5},
{args{5, 5}, 10},
{args{6, 5}, 10},
{args{13, 5}, 15},
{args{95, 5}, 100},
}
for _, tt := range tests {
got := rangeForTimestamp(tt.args.t, tt.args.width)
require.Equal(t, tt.expected, got)
}
}
// TestChunkReader_ConcurrentReads checks that the chunk result can be read concurrently.
// Regression test for https://github.com/prometheus/prometheus/pull/6514.
func TestChunkReader_ConcurrentReads(t *testing.T) {
chks := []chunks.Meta{
tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 1}}),
tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 2}}),
tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 3}}),
tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 4}}),
tsdbutil.ChunkFromSamples([]tsdbutil.Sample{sample{1, 5}}),
}
tempDir := t.TempDir()
chunkw, err := chunks.NewWriter(tempDir)
require.NoError(t, err)
require.NoError(t, chunkw.WriteChunks(chks...))
require.NoError(t, chunkw.Close())
r, err := chunks.NewDirReader(tempDir, nil)
require.NoError(t, err)
var wg sync.WaitGroup
for _, chk := range chks {
for i := 0; i < 100; i++ {
wg.Add(1)
go func(chunk chunks.Meta) {
defer wg.Done()
chkAct, err := r.Chunk(chunk.Ref)
require.NoError(t, err)
require.Equal(t, chunk.Chunk.Bytes(), chkAct.Bytes())
}(chk)
}
wg.Wait()
}
require.NoError(t, r.Close())
}
2020-06-12 03:29:26 -07:00
// TestCompactHead ensures that the head compaction
// creates a block that is ready for loading and
// does not cause data loss.
// This test:
// * opens a storage;
// * appends values;
// * compacts the head; and
// * queries the db to ensure the samples are present from the compacted head.
func TestCompactHead(t *testing.T) {
dbDir := t.TempDir()
2020-06-12 03:29:26 -07:00
// Open a DB and append data to the WAL.
tsdbCfg := &Options{
RetentionDuration: int64(time.Hour * 24 * 15 / time.Millisecond),
NoLockfile: true,
MinBlockDuration: int64(time.Hour * 2 / time.Millisecond),
MaxBlockDuration: int64(time.Hour * 2 / time.Millisecond),
WALCompression: true,
}
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(dbDir, log.NewNopLogger(), prometheus.NewRegistry(), tsdbCfg, nil)
require.NoError(t, err)
ctx := context.Background()
app := db.Appender(ctx)
2020-06-12 03:29:26 -07:00
var expSamples []sample
maxt := 100
for i := 0; i < maxt; i++ {
val := rand.Float64()
_, err := app.Append(0, labels.FromStrings("a", "b"), int64(i), val)
require.NoError(t, err)
2020-06-12 03:29:26 -07:00
expSamples = append(expSamples, sample{int64(i), val})
}
require.NoError(t, app.Commit())
2020-06-12 03:29:26 -07:00
// Compact the Head to create a new block.
require.NoError(t, db.CompactHead(NewRangeHead(db.Head(), 0, int64(maxt)-1)))
require.NoError(t, db.Close())
2020-06-12 03:29:26 -07:00
// Delete everything but the new block and
// reopen the db to query it to ensure it includes the head data.
require.NoError(t, deleteNonBlocks(db.Dir()))
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err = Open(dbDir, log.NewNopLogger(), prometheus.NewRegistry(), tsdbCfg, nil)
require.NoError(t, err)
require.Equal(t, 1, len(db.Blocks()))
require.Equal(t, int64(maxt), db.Head().MinTime())
defer func() { require.NoError(t, db.Close()) }()
2020-06-12 03:29:26 -07:00
querier, err := db.Querier(context.Background(), 0, int64(maxt)-1)
require.NoError(t, err)
defer func() { require.NoError(t, querier.Close()) }()
2020-06-12 03:29:26 -07:00
seriesSet := querier.Select(false, nil, &labels.Matcher{Type: labels.MatchEqual, Name: "a", Value: "b"})
2020-06-12 03:29:26 -07:00
var actSamples []sample
for seriesSet.Next() {
series := seriesSet.At().Iterator()
for series.Next() {
time, val := series.At()
actSamples = append(actSamples, sample{int64(time), val})
}
require.NoError(t, series.Err())
2020-06-12 03:29:26 -07:00
}
require.Equal(t, expSamples, actSamples)
require.NoError(t, seriesSet.Err())
2020-06-12 03:29:26 -07:00
}
func deleteNonBlocks(dbDir string) error {
dirs, err := os.ReadDir(dbDir)
2020-06-12 03:29:26 -07:00
if err != nil {
return err
}
for _, dir := range dirs {
if ok := isBlockDir(dir); !ok {
if err := os.RemoveAll(filepath.Join(dbDir, dir.Name())); err != nil {
return err
}
}
}
dirs, err = os.ReadDir(dbDir)
2020-06-12 03:29:26 -07:00
if err != nil {
return err
}
for _, dir := range dirs {
if ok := isBlockDir(dir); !ok {
return errors.Errorf("root folder:%v still hase non block directory:%v", dbDir, dir.Name())
}
}
return nil
}
func TestOpen_VariousBlockStates(t *testing.T) {
tmpDir := t.TempDir()
var (
expectedLoadedDirs = map[string]struct{}{}
expectedRemovedDirs = map[string]struct{}{}
expectedIgnoredDirs = map[string]struct{}{}
)
{
// Ok blocks; should be loaded.
expectedLoadedDirs[createBlock(t, tmpDir, genSeries(10, 2, 0, 10))] = struct{}{}
expectedLoadedDirs[createBlock(t, tmpDir, genSeries(10, 2, 10, 20))] = struct{}{}
}
{
// Block to repair; should be repaired & loaded.
dbDir := filepath.Join("testdata", "repair_index_version", "01BZJ9WJQPWHGNC2W4J9TA62KC")
outDir := filepath.Join(tmpDir, "01BZJ9WJQPWHGNC2W4J9TA62KC")
expectedLoadedDirs[outDir] = struct{}{}
// Touch chunks dir in block.
require.NoError(t, os.MkdirAll(filepath.Join(dbDir, "chunks"), 0o777))
defer func() {
require.NoError(t, os.RemoveAll(filepath.Join(dbDir, "chunks")))
}()
require.NoError(t, os.Mkdir(outDir, os.ModePerm))
require.NoError(t, fileutil.CopyDirs(dbDir, outDir))
}
{
// Missing meta.json; should be ignored and only logged.
// TODO(bwplotka): Probably add metric.
dir := createBlock(t, tmpDir, genSeries(10, 2, 20, 30))
expectedIgnoredDirs[dir] = struct{}{}
require.NoError(t, os.Remove(filepath.Join(dir, metaFilename)))
}
{
// Tmp blocks during creation; those should be removed on start.
dir := createBlock(t, tmpDir, genSeries(10, 2, 30, 40))
require.NoError(t, fileutil.Replace(dir, dir+tmpForCreationBlockDirSuffix))
expectedRemovedDirs[dir+tmpForCreationBlockDirSuffix] = struct{}{}
// Tmp blocks during deletion; those should be removed on start.
dir = createBlock(t, tmpDir, genSeries(10, 2, 40, 50))
require.NoError(t, fileutil.Replace(dir, dir+tmpForDeletionBlockDirSuffix))
expectedRemovedDirs[dir+tmpForDeletionBlockDirSuffix] = struct{}{}
// Pre-2.21 tmp blocks; those should be removed on start.
dir = createBlock(t, tmpDir, genSeries(10, 2, 50, 60))
require.NoError(t, fileutil.Replace(dir, dir+tmpLegacy))
expectedRemovedDirs[dir+tmpLegacy] = struct{}{}
}
{
// One ok block; but two should be replaced.
dir := createBlock(t, tmpDir, genSeries(10, 2, 50, 60))
expectedLoadedDirs[dir] = struct{}{}
m, _, err := readMetaFile(dir)
require.NoError(t, err)
compacted := createBlock(t, tmpDir, genSeries(10, 2, 50, 55))
expectedRemovedDirs[compacted] = struct{}{}
m.Compaction.Parents = append(m.Compaction.Parents,
BlockDesc{ULID: ulid.MustParse(filepath.Base(compacted))},
BlockDesc{ULID: ulid.MustNew(1, nil)},
BlockDesc{ULID: ulid.MustNew(123, nil)},
)
// Regression test: Already removed parent can be still in list, which was causing Open errors.
m.Compaction.Parents = append(m.Compaction.Parents, BlockDesc{ULID: ulid.MustParse(filepath.Base(compacted))})
m.Compaction.Parents = append(m.Compaction.Parents, BlockDesc{ULID: ulid.MustParse(filepath.Base(compacted))})
_, err = writeMetaFile(log.NewLogfmtLogger(os.Stderr), dir, m)
require.NoError(t, err)
}
tmpCheckpointDir := path.Join(tmpDir, "wal/checkpoint.00000001.tmp")
err := os.MkdirAll(tmpCheckpointDir, 0o777)
require.NoError(t, err)
opts := DefaultOptions()
opts.RetentionDuration = 0
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(tmpDir, log.NewLogfmtLogger(os.Stderr), nil, opts, nil)
require.NoError(t, err)
loadedBlocks := db.Blocks()
var loaded int
for _, l := range loadedBlocks {
if _, ok := expectedLoadedDirs[filepath.Join(tmpDir, l.meta.ULID.String())]; !ok {
t.Fatal("unexpected block", l.meta.ULID, "was loaded")
}
loaded++
}
require.Equal(t, len(expectedLoadedDirs), loaded)
require.NoError(t, db.Close())
files, err := os.ReadDir(tmpDir)
require.NoError(t, err)
var ignored int
for _, f := range files {
if _, ok := expectedRemovedDirs[filepath.Join(tmpDir, f.Name())]; ok {
t.Fatal("expected", filepath.Join(tmpDir, f.Name()), "to be removed, but still exists")
}
if _, ok := expectedIgnoredDirs[filepath.Join(tmpDir, f.Name())]; ok {
ignored++
}
}
require.Equal(t, len(expectedIgnoredDirs), ignored)
_, err = os.Stat(tmpCheckpointDir)
require.True(t, os.IsNotExist(err))
}
func TestOneCheckpointPerCompactCall(t *testing.T) {
blockRange := int64(1000)
tsdbCfg := &Options{
RetentionDuration: blockRange * 1000,
NoLockfile: true,
MinBlockDuration: blockRange,
MaxBlockDuration: blockRange,
}
tmpDir := t.TempDir()
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err := Open(tmpDir, log.NewNopLogger(), prometheus.NewRegistry(), tsdbCfg, nil)
require.NoError(t, err)
t.Cleanup(func() {
require.NoError(t, db.Close())
})
db.DisableCompactions()
// Case 1: Lot's of uncompacted data in Head.
lbls := labels.FromStrings("foo_d", "choco_bar")
// Append samples spanning 59 block ranges.
app := db.Appender(context.Background())
for i := int64(0); i < 60; i++ {
_, err := app.Append(0, lbls, blockRange*i, rand.Float64())
require.NoError(t, err)
_, err = app.Append(0, lbls, (blockRange*i)+blockRange/2, rand.Float64())
require.NoError(t, err)
// Rotate the WAL file so that there is >3 files for checkpoint to happen.
require.NoError(t, db.head.wal.NextSegment())
}
require.NoError(t, app.Commit())
// Check the existing WAL files.
first, last, err := wal.Segments(db.head.wal.Dir())
require.NoError(t, err)
require.Equal(t, 0, first)
require.Equal(t, 60, last)
require.Equal(t, 0.0, prom_testutil.ToFloat64(db.head.metrics.checkpointCreationTotal))
require.NoError(t, db.Compact())
require.Equal(t, 1.0, prom_testutil.ToFloat64(db.head.metrics.checkpointCreationTotal))
// As the data spans for 59 blocks, 58 go to disk and 1 remains in Head.
require.Equal(t, 58, len(db.Blocks()))
// Though WAL was truncated only once, head should be truncated after each compaction.
require.Equal(t, 58.0, prom_testutil.ToFloat64(db.head.metrics.headTruncateTotal))
// The compaction should have only truncated first 2/3 of WAL (while also rotating the files).
first, last, err = wal.Segments(db.head.wal.Dir())
require.NoError(t, err)
require.Equal(t, 40, first)
require.Equal(t, 61, last)
// The first checkpoint would be for first 2/3rd of WAL, hence till 39.
// That should be the last checkpoint.
_, cno, err := wal.LastCheckpoint(db.head.wal.Dir())
require.NoError(t, err)
require.Equal(t, 39, cno)
// Case 2: Old blocks on disk.
// The above blocks will act as old blocks.
// Creating a block to cover the data in the Head so that
// Head will skip the data during replay and start fresh.
blocks := db.Blocks()
newBlockMint := blocks[len(blocks)-1].Meta().MaxTime
newBlockMaxt := db.Head().MaxTime() + 1
require.NoError(t, db.Close())
createBlock(t, db.dir, genSeries(1, 1, newBlockMint, newBlockMaxt))
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
db, err = Open(db.dir, log.NewNopLogger(), prometheus.NewRegistry(), tsdbCfg, nil)
require.NoError(t, err)
db.DisableCompactions()
// 1 block more.
require.Equal(t, 59, len(db.Blocks()))
// No series in Head because of this new block.
require.Equal(t, 0, int(db.head.NumSeries()))
// Adding sample way into the future.
app = db.Appender(context.Background())
_, err = app.Append(0, lbls, blockRange*120, rand.Float64())
require.NoError(t, err)
require.NoError(t, app.Commit())
// The mint of head is the last block maxt, that means the gap between mint and maxt
// of Head is too large. This will trigger many compactions.
require.Equal(t, newBlockMaxt, db.head.MinTime())
// Another WAL file was rotated.
first, last, err = wal.Segments(db.head.wal.Dir())
require.NoError(t, err)
require.Equal(t, 40, first)
require.Equal(t, 62, last)
require.Equal(t, 0.0, prom_testutil.ToFloat64(db.head.metrics.checkpointCreationTotal))
require.NoError(t, db.Compact())
require.Equal(t, 1.0, prom_testutil.ToFloat64(db.head.metrics.checkpointCreationTotal))
// No new blocks should be created as there was not data in between the new samples and the blocks.
require.Equal(t, 59, len(db.Blocks()))
// The compaction should have only truncated first 2/3 of WAL (while also rotating the files).
first, last, err = wal.Segments(db.head.wal.Dir())
require.NoError(t, err)
require.Equal(t, 55, first)
require.Equal(t, 63, last)
// The first checkpoint would be for first 2/3rd of WAL, hence till 54.
// That should be the last checkpoint.
_, cno, err = wal.LastCheckpoint(db.head.wal.Dir())
require.NoError(t, err)
require.Equal(t, 54, cno)
}
func TestNoPanicOnTSDBOpenError(t *testing.T) {
tmpdir := t.TempDir()
// Taking the lock will cause a TSDB startup error.
l, err := tsdbutil.NewDirLocker(tmpdir, "tsdb", log.NewNopLogger(), nil)
require.NoError(t, err)
require.NoError(t, l.Lock())
React UI: Add Starting Screen (#8662) * Added walreplay API endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added starting page to react-ui Signed-off-by: Levi Harrison <git@leviharrison.dev> * Documented the new endpoint Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed typos Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com> * Removed logo Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed isResponding to isUnexpected Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed width of progress bar Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added DB stats object Signed-off-by: Levi Harrison <git@leviharrison.dev> * Updated starting page to work with new fields Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 2) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 3) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 4) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (and also implementing a method this time) (pt. 5) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed const to let Signed-off-by: Levi Harrison <git@leviharrison.dev> * Passing nil (pt. 6) Signed-off-by: Levi Harrison <git@leviharrison.dev> * Remove SetStats method Signed-off-by: Levi Harrison <git@leviharrison.dev> * Added comma Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed api Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed to triple equals Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed data response types Signed-off-by: Levi Harrison <git@leviharrison.dev> * Don't return pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Changed version Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed interface issue Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed pointer Signed-off-by: Levi Harrison <git@leviharrison.dev> * Fixed copying lock value error Signed-off-by: Levi Harrison <git@leviharrison.dev> Co-authored-by: Julius Volz <julius.volz@gmail.com>
2021-06-05 07:29:32 -07:00
_, err = Open(tmpdir, nil, nil, DefaultOptions(), nil)
require.Error(t, err)
require.NoError(t, l.Release())
}
func TestLockfile(t *testing.T) {
tsdbutil.TestDirLockerUsage(t, func(t *testing.T, data string, createLock bool) (*tsdbutil.DirLocker, testutil.Closer) {
opts := DefaultOptions()
opts.NoLockfile = !createLock
// Create the DB. This should create lockfile and its metrics.
db, err := Open(data, nil, nil, opts, nil)
require.NoError(t, err)
return db.locker, testutil.NewCallbackCloser(func() {
require.NoError(t, db.Close())
})
})
}
func TestQuerier_ShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks(t *testing.T) {
t.Skip("TODO: investigate why process crash in CI")
const numRuns = 5
for i := 1; i <= numRuns; i++ {
t.Run(strconv.Itoa(i), func(t *testing.T) {
testQuerierShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks(t)
})
}
}
func testQuerierShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks(t *testing.T) {
const (
numSeries = 1000
numStressIterations = 10000
minStressAllocationBytes = 128 * 1024
maxStressAllocationBytes = 512 * 1024
)
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
// Disable compactions so we can control it.
db.DisableCompactions()
// Generate the metrics we're going to append.
metrics := make([]labels.Labels, 0, numSeries)
for i := 0; i < numSeries; i++ {
metrics = append(metrics, labels.FromStrings(labels.MetricName, fmt.Sprintf("test_%d", i)))
}
// Push 1 sample every 15s for 2x the block duration period.
ctx := context.Background()
interval := int64(15 * time.Second / time.Millisecond)
ts := int64(0)
for ; ts < 2*DefaultBlockDuration; ts += interval {
app := db.Appender(ctx)
for _, metric := range metrics {
_, err := app.Append(0, metric, ts, float64(ts))
require.NoError(t, err)
}
require.NoError(t, app.Commit())
}
// Compact the TSDB head for the first time. We expect the head chunks file has been cut.
require.NoError(t, db.Compact())
require.Equal(t, float64(1), prom_testutil.ToFloat64(db.Head().metrics.headTruncateTotal))
// Push more samples for another 1x block duration period.
for ; ts < 3*DefaultBlockDuration; ts += interval {
app := db.Appender(ctx)
for _, metric := range metrics {
_, err := app.Append(0, metric, ts, float64(ts))
require.NoError(t, err)
}
require.NoError(t, app.Commit())
}
// At this point we expect 2 mmap-ed head chunks.
// Get a querier and make sure it's closed only once the test is over.
querier, err := db.Querier(ctx, 0, math.MaxInt64)
require.NoError(t, err)
defer func() {
require.NoError(t, querier.Close())
}()
// Query back all series.
hints := &storage.SelectHints{Start: 0, End: math.MaxInt64, Step: interval}
seriesSet := querier.Select(true, hints, labels.MustNewMatcher(labels.MatchRegexp, labels.MetricName, ".+"))
// Fetch samples iterators from all series.
var iterators []chunkenc.Iterator
actualSeries := 0
for seriesSet.Next() {
actualSeries++
// Get the iterator and call Next() so that we're sure the chunk is loaded.
it := seriesSet.At().Iterator()
it.Next()
it.At()
iterators = append(iterators, it)
}
require.NoError(t, seriesSet.Err())
require.Equal(t, actualSeries, numSeries)
// Compact the TSDB head again.
require.NoError(t, db.Compact())
require.Equal(t, float64(2), prom_testutil.ToFloat64(db.Head().metrics.headTruncateTotal))
// At this point we expect 1 head chunk has been deleted.
// Stress the memory and call GC. This is required to increase the chances
// the chunk memory area is released to the kernel.
var buf []byte
for i := 0; i < numStressIterations; i++ {
//nolint:staticcheck
buf = append(buf, make([]byte, minStressAllocationBytes+rand.Int31n(maxStressAllocationBytes-minStressAllocationBytes))...)
if i%1000 == 0 {
buf = nil
}
}
// Iterate samples. Here we're summing it just to make sure no golang compiler
// optimization triggers in case we discard the result of it.At().
var sum float64
var firstErr error
for _, it := range iterators {
for it.Next() {
_, v := it.At()
sum += v
}
if err := it.Err(); err != nil {
firstErr = err
}
}
// After having iterated all samples we also want to be sure no error occurred or
// the "cannot populate chunk XXX: not found" error occurred. This error can occur
// when the iterator tries to fetch an head chunk which has been offloaded because
// of the head compaction in the meanwhile.
if firstErr != nil && !strings.Contains(firstErr.Error(), "cannot populate chunk") {
t.Fatalf("unexpected error: %s", firstErr.Error())
}
}
func TestChunkQuerier_ShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks(t *testing.T) {
t.Skip("TODO: investigate why process crash in CI")
const numRuns = 5
for i := 1; i <= numRuns; i++ {
t.Run(strconv.Itoa(i), func(t *testing.T) {
testChunkQuerierShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks(t)
})
}
}
func testChunkQuerierShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks(t *testing.T) {
const (
numSeries = 1000
numStressIterations = 10000
minStressAllocationBytes = 128 * 1024
maxStressAllocationBytes = 512 * 1024
)
db := openTestDB(t, nil, nil)
defer func() {
require.NoError(t, db.Close())
}()
// Disable compactions so we can control it.
db.DisableCompactions()
// Generate the metrics we're going to append.
metrics := make([]labels.Labels, 0, numSeries)
for i := 0; i < numSeries; i++ {
metrics = append(metrics, labels.FromStrings(labels.MetricName, fmt.Sprintf("test_%d", i)))
}
// Push 1 sample every 15s for 2x the block duration period.
ctx := context.Background()
interval := int64(15 * time.Second / time.Millisecond)
ts := int64(0)
for ; ts < 2*DefaultBlockDuration; ts += interval {
app := db.Appender(ctx)
for _, metric := range metrics {
_, err := app.Append(0, metric, ts, float64(ts))
require.NoError(t, err)
}
require.NoError(t, app.Commit())
}
// Compact the TSDB head for the first time. We expect the head chunks file has been cut.
require.NoError(t, db.Compact())
require.Equal(t, float64(1), prom_testutil.ToFloat64(db.Head().metrics.headTruncateTotal))
// Push more samples for another 1x block duration period.
for ; ts < 3*DefaultBlockDuration; ts += interval {
app := db.Appender(ctx)
for _, metric := range metrics {
_, err := app.Append(0, metric, ts, float64(ts))
require.NoError(t, err)
}
require.NoError(t, app.Commit())
}
// At this point we expect 2 mmap-ed head chunks.
// Get a querier and make sure it's closed only once the test is over.
querier, err := db.ChunkQuerier(ctx, 0, math.MaxInt64)
require.NoError(t, err)
defer func() {
require.NoError(t, querier.Close())
}()
// Query back all series.
hints := &storage.SelectHints{Start: 0, End: math.MaxInt64, Step: interval}
seriesSet := querier.Select(true, hints, labels.MustNewMatcher(labels.MatchRegexp, labels.MetricName, ".+"))
// Iterate all series and get their chunks.
var chunks []chunkenc.Chunk
actualSeries := 0
for seriesSet.Next() {
actualSeries++
for it := seriesSet.At().Iterator(); it.Next(); {
chunks = append(chunks, it.At().Chunk)
}
}
require.NoError(t, seriesSet.Err())
require.Equal(t, actualSeries, numSeries)
// Compact the TSDB head again.
require.NoError(t, db.Compact())
require.Equal(t, float64(2), prom_testutil.ToFloat64(db.Head().metrics.headTruncateTotal))
// At this point we expect 1 head chunk has been deleted.
// Stress the memory and call GC. This is required to increase the chances
// the chunk memory area is released to the kernel.
var buf []byte
for i := 0; i < numStressIterations; i++ {
//nolint:staticcheck
buf = append(buf, make([]byte, minStressAllocationBytes+rand.Int31n(maxStressAllocationBytes-minStressAllocationBytes))...)
if i%1000 == 0 {
buf = nil
}
}
// Iterate chunks and read their bytes slice. Here we're computing the CRC32
// just to iterate through the bytes slice. We don't really care the reason why
// we read this data, we just need to read it to make sure the memory address
// of the []byte is still valid.
chkCRC32 := newCRC32()
for _, chunk := range chunks {
chkCRC32.Reset()
_, err := chkCRC32.Write(chunk.Bytes())
require.NoError(t, err)
}
}
func newTestDB(t *testing.T) *DB {
dir := t.TempDir()
db, err := Open(dir, nil, nil, DefaultOptions(), nil)
require.NoError(t, err)
t.Cleanup(func() {
require.NoError(t, db.Close())
})
return db
}
// Tests https://github.com/prometheus/prometheus/issues/10291#issuecomment-1044373110.
func TestDBPanicOnMmappingHeadChunk(t *testing.T) {
dir := t.TempDir()
db, err := Open(dir, nil, nil, DefaultOptions(), nil)
require.NoError(t, err)
db.DisableCompactions()
// Choosing scrape interval of 45s to have chunk larger than 1h.
itvl := int64(45 * time.Second / time.Millisecond)
lastTs := int64(0)
addSamples := func(numSamples int) {
app := db.Appender(context.Background())
var ref storage.SeriesRef
lbls := labels.FromStrings("__name__", "testing", "foo", "bar")
for i := 0; i < numSamples; i++ {
ref, err = app.Append(ref, lbls, lastTs, float64(lastTs))
require.NoError(t, err)
lastTs += itvl
if i%10 == 0 {
require.NoError(t, app.Commit())
app = db.Appender(context.Background())
}
}
require.NoError(t, app.Commit())
}
// Ingest samples upto 2h50m to make the head "about to compact".
numSamples := int(170*time.Minute/time.Millisecond) / int(itvl)
addSamples(numSamples)
require.Len(t, db.Blocks(), 0)
require.NoError(t, db.Compact())
require.Len(t, db.Blocks(), 0)
// Restarting.
require.NoError(t, db.Close())
db, err = Open(dir, nil, nil, DefaultOptions(), nil)
require.NoError(t, err)
db.DisableCompactions()
// Ingest samples upto 20m more to make the head compact.
numSamples = int(20*time.Minute/time.Millisecond) / int(itvl)
addSamples(numSamples)
require.Len(t, db.Blocks(), 0)
require.NoError(t, db.Compact())
require.Len(t, db.Blocks(), 1)
// More samples to m-map and panic.
numSamples = int(120*time.Minute/time.Millisecond) / int(itvl)
addSamples(numSamples)
require.NoError(t, db.Close())
}
Introduce TSDB changes for appending metadata to the WAL (#10972) * Append metadata to the WAL Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove extra whitespace; Reword some docstrings and comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use RLock() for hasNewMetadata check Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use single byte for metric type in RefMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update proposed WAL format for single-byte type metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implementa MetadataAppender interface for the Agent Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of review comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Amend description of metadata in wal.md Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Correct key used to retrieve metadata from cache When we're setting metadata entries in the scrapeCace, we're using the p.Help(), p.Unit(), p.Type() helpers, which retrieve the series name and use it as the cache key. When checking for cache entries though, we used p.Series() as the key, which included the metric name _with_ its labels. That meant that we were never actually hitting the cache. We're fixing this by utiling the __name__ internal label for correctly getting the cache entries after they've been set by setHelp(), setType() or setUnit(). Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Put feature behind a feature flag Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix AppendMetadata docstring Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reorder WAL format document Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Change error message of AppendMetadata; Fix access of s.meta in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse temporary buffer in Metadata encoder Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Only keep latest metadata for each refID during checkpointing Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test that's referencing decoding metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Avoid creating metadata block if no new metadata are present Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for corrupt metadata block and relevant record type Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix CR comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Extract logic about changing metadata in an anonymous function Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implement new proposed WAL format and amend relevant tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use 'const' for metadata field names Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Apply metadata to head memSeries in Commit, not in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add docstring and rename extracted helper in scrape.go Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for tsdb-related cases Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol1 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol2 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix Windows test by closing WAL reader files Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use switch instead of two if statements in metadata decoding Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments around TestMetadata* tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add code for replaying WAL; test correctness of in-memory data after a replay Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove scrape-loop related code from PR Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify tests by sorting slices before comparison Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test to use separate transactions Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Empty out buffer and record slices after encoding latest metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linting issue Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update calculation for DroppedMetadata metric Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Rename MetadataAppender interface and AppendMetadata method to MetadataUpdater/UpdateMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse buffer when encoding latest metadata for each series Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments; Check all returned error values using two helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify use of helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Satisfy linter Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com>
2022-07-19 01:58:52 -07:00
func TestMetadataInWAL(t *testing.T) {
updateMetadata := func(t *testing.T, app storage.Appender, s labels.Labels, m metadata.Metadata) {
_, err := app.UpdateMetadata(0, s, m)
require.NoError(t, err)
}
db := newTestDB(t)
ctx := context.Background()
// Add some series so we can append metadata to them.
app := db.Appender(ctx)
s1 := labels.FromStrings("a", "b")
s2 := labels.FromStrings("c", "d")
s3 := labels.FromStrings("e", "f")
s4 := labels.FromStrings("g", "h")
for _, s := range []labels.Labels{s1, s2, s3, s4} {
_, err := app.Append(0, s, 0, 0)
require.NoError(t, err)
}
require.NoError(t, app.Commit())
// Add a first round of metadata to the first three series.
// Re-take the Appender, as the previous Commit will have it closed.
m1 := metadata.Metadata{Type: "gauge", Unit: "unit_1", Help: "help_1"}
m2 := metadata.Metadata{Type: "gauge", Unit: "unit_2", Help: "help_2"}
m3 := metadata.Metadata{Type: "gauge", Unit: "unit_3", Help: "help_3"}
app = db.Appender(ctx)
updateMetadata(t, app, s1, m1)
updateMetadata(t, app, s2, m2)
updateMetadata(t, app, s3, m3)
require.NoError(t, app.Commit())
// Add a replicated metadata entry to the first series,
// a completely new metadata entry for the fourth series,
// and a changed metadata entry to the second series.
m4 := metadata.Metadata{Type: "counter", Unit: "unit_4", Help: "help_4"}
m5 := metadata.Metadata{Type: "counter", Unit: "unit_5", Help: "help_5"}
app = db.Appender(ctx)
updateMetadata(t, app, s1, m1)
updateMetadata(t, app, s4, m4)
updateMetadata(t, app, s2, m5)
require.NoError(t, app.Commit())
// Read the WAL to see if the disk storage format is correct.
recs := readTestWAL(t, path.Join(db.Dir(), "wal"))
var gotMetadataBlocks [][]record.RefMetadata
for _, rec := range recs {
if mr, ok := rec.([]record.RefMetadata); ok {
gotMetadataBlocks = append(gotMetadataBlocks, mr)
}
}
expectedMetadata := []record.RefMetadata{
{Ref: 1, Type: record.GetMetricType(m1.Type), Unit: m1.Unit, Help: m1.Help},
{Ref: 2, Type: record.GetMetricType(m2.Type), Unit: m2.Unit, Help: m2.Help},
{Ref: 3, Type: record.GetMetricType(m3.Type), Unit: m3.Unit, Help: m3.Help},
{Ref: 4, Type: record.GetMetricType(m4.Type), Unit: m4.Unit, Help: m4.Help},
{Ref: 2, Type: record.GetMetricType(m5.Type), Unit: m5.Unit, Help: m5.Help},
}
require.Len(t, gotMetadataBlocks, 2)
require.Equal(t, expectedMetadata[:3], gotMetadataBlocks[0])
require.Equal(t, expectedMetadata[3:], gotMetadataBlocks[1])
}
func TestMetadataCheckpointingOnlyKeepsLatestEntry(t *testing.T) {
updateMetadata := func(t *testing.T, app storage.Appender, s labels.Labels, m metadata.Metadata) {
_, err := app.UpdateMetadata(0, s, m)
require.NoError(t, err)
}
ctx := context.Background()
numSamples := 10000
hb, w := newTestHead(t, int64(numSamples)*10, false)
// Add some series so we can append metadata to them.
app := hb.Appender(ctx)
s1 := labels.FromStrings("a", "b")
s2 := labels.FromStrings("c", "d")
s3 := labels.FromStrings("e", "f")
s4 := labels.FromStrings("g", "h")
for _, s := range []labels.Labels{s1, s2, s3, s4} {
_, err := app.Append(0, s, 0, 0)
require.NoError(t, err)
}
require.NoError(t, app.Commit())
// Add a first round of metadata to the first three series.
// Re-take the Appender, as the previous Commit will have it closed.
m1 := metadata.Metadata{Type: "gauge", Unit: "unit_1", Help: "help_1"}
m2 := metadata.Metadata{Type: "gauge", Unit: "unit_2", Help: "help_2"}
m3 := metadata.Metadata{Type: "gauge", Unit: "unit_3", Help: "help_3"}
m4 := metadata.Metadata{Type: "gauge", Unit: "unit_4", Help: "help_4"}
app = hb.Appender(ctx)
updateMetadata(t, app, s1, m1)
updateMetadata(t, app, s2, m2)
updateMetadata(t, app, s3, m3)
updateMetadata(t, app, s4, m4)
require.NoError(t, app.Commit())
// Update metadata for first series.
m5 := metadata.Metadata{Type: "counter", Unit: "unit_5", Help: "help_5"}
app = hb.Appender(ctx)
updateMetadata(t, app, s1, m5)
require.NoError(t, app.Commit())
// Switch back-and-forth metadata for second series.
// Since it ended on a new metadata record, we expect a single new entry.
m6 := metadata.Metadata{Type: "counter", Unit: "unit_6", Help: "help_6"}
app = hb.Appender(ctx)
updateMetadata(t, app, s2, m6)
require.NoError(t, app.Commit())
app = hb.Appender(ctx)
updateMetadata(t, app, s2, m2)
require.NoError(t, app.Commit())
app = hb.Appender(ctx)
updateMetadata(t, app, s2, m6)
require.NoError(t, app.Commit())
app = hb.Appender(ctx)
updateMetadata(t, app, s2, m2)
require.NoError(t, app.Commit())
app = hb.Appender(ctx)
updateMetadata(t, app, s2, m6)
require.NoError(t, app.Commit())
// Let's create a checkpoint.
first, last, err := wal.Segments(w.Dir())
require.NoError(t, err)
keep := func(id chunks.HeadSeriesRef) bool {
return id != 3
}
_, err = wal.Checkpoint(log.NewNopLogger(), w, first, last-1, keep, 0)
require.NoError(t, err)
// Confirm there's been a checkpoint.
cdir, _, err := wal.LastCheckpoint(w.Dir())
require.NoError(t, err)
// Read in checkpoint and WAL.
recs := readTestWAL(t, cdir)
var gotMetadataBlocks [][]record.RefMetadata
for _, rec := range recs {
if mr, ok := rec.([]record.RefMetadata); ok {
gotMetadataBlocks = append(gotMetadataBlocks, mr)
}
}
// There should only be 1 metadata block present, with only the latest
// metadata kept around.
wantMetadata := []record.RefMetadata{
{Ref: 1, Type: record.GetMetricType(m5.Type), Unit: m5.Unit, Help: m5.Help},
{Ref: 2, Type: record.GetMetricType(m6.Type), Unit: m6.Unit, Help: m6.Help},
{Ref: 4, Type: record.GetMetricType(m4.Type), Unit: m4.Unit, Help: m4.Help},
}
require.Len(t, gotMetadataBlocks, 1)
require.Len(t, gotMetadataBlocks[0], 3)
gotMetadataBlock := gotMetadataBlocks[0]
sort.Slice(gotMetadataBlock, func(i, j int) bool { return gotMetadataBlock[i].Ref < gotMetadataBlock[j].Ref })
require.Equal(t, wantMetadata, gotMetadataBlock)
require.NoError(t, hb.Close())
}
func TestMetadataAssertInMemoryData(t *testing.T) {
updateMetadata := func(t *testing.T, app storage.Appender, s labels.Labels, m metadata.Metadata) {
_, err := app.UpdateMetadata(0, s, m)
require.NoError(t, err)
}
db := openTestDB(t, nil, nil)
ctx := context.Background()
// Add some series so we can append metadata to them.
app := db.Appender(ctx)
s1 := labels.FromStrings("a", "b")
s2 := labels.FromStrings("c", "d")
s3 := labels.FromStrings("e", "f")
s4 := labels.FromStrings("g", "h")
for _, s := range []labels.Labels{s1, s2, s3, s4} {
_, err := app.Append(0, s, 0, 0)
require.NoError(t, err)
}
require.NoError(t, app.Commit())
// Add a first round of metadata to the first three series.
// The in-memory data held in the db Head should hold the metadata.
m1 := metadata.Metadata{Type: "gauge", Unit: "unit_1", Help: "help_1"}
m2 := metadata.Metadata{Type: "gauge", Unit: "unit_2", Help: "help_2"}
m3 := metadata.Metadata{Type: "gauge", Unit: "unit_3", Help: "help_3"}
app = db.Appender(ctx)
updateMetadata(t, app, s1, m1)
updateMetadata(t, app, s2, m2)
updateMetadata(t, app, s3, m3)
require.NoError(t, app.Commit())
series1 := db.head.series.getByHash(s1.Hash(), s1)
series2 := db.head.series.getByHash(s2.Hash(), s2)
series3 := db.head.series.getByHash(s3.Hash(), s3)
series4 := db.head.series.getByHash(s4.Hash(), s4)
require.Equal(t, *series1.meta, m1)
require.Equal(t, *series2.meta, m2)
require.Equal(t, *series3.meta, m3)
require.Nil(t, series4.meta)
Introduce TSDB changes for appending metadata to the WAL (#10972) * Append metadata to the WAL Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove extra whitespace; Reword some docstrings and comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use RLock() for hasNewMetadata check Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use single byte for metric type in RefMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update proposed WAL format for single-byte type metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implementa MetadataAppender interface for the Agent Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of review comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Amend description of metadata in wal.md Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Correct key used to retrieve metadata from cache When we're setting metadata entries in the scrapeCace, we're using the p.Help(), p.Unit(), p.Type() helpers, which retrieve the series name and use it as the cache key. When checking for cache entries though, we used p.Series() as the key, which included the metric name _with_ its labels. That meant that we were never actually hitting the cache. We're fixing this by utiling the __name__ internal label for correctly getting the cache entries after they've been set by setHelp(), setType() or setUnit(). Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Put feature behind a feature flag Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix AppendMetadata docstring Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reorder WAL format document Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Change error message of AppendMetadata; Fix access of s.meta in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse temporary buffer in Metadata encoder Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Only keep latest metadata for each refID during checkpointing Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test that's referencing decoding metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Avoid creating metadata block if no new metadata are present Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for corrupt metadata block and relevant record type Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix CR comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Extract logic about changing metadata in an anonymous function Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implement new proposed WAL format and amend relevant tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use 'const' for metadata field names Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Apply metadata to head memSeries in Commit, not in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add docstring and rename extracted helper in scrape.go Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for tsdb-related cases Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol1 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol2 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix Windows test by closing WAL reader files Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use switch instead of two if statements in metadata decoding Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments around TestMetadata* tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add code for replaying WAL; test correctness of in-memory data after a replay Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove scrape-loop related code from PR Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify tests by sorting slices before comparison Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test to use separate transactions Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Empty out buffer and record slices after encoding latest metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linting issue Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update calculation for DroppedMetadata metric Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Rename MetadataAppender interface and AppendMetadata method to MetadataUpdater/UpdateMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse buffer when encoding latest metadata for each series Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments; Check all returned error values using two helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify use of helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Satisfy linter Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com>
2022-07-19 01:58:52 -07:00
// Add a replicated metadata entry to the first series,
// a changed metadata entry to the second series,
// and a completely new metadata entry for the fourth series.
// The in-memory data held in the db Head should be correctly updated.
m4 := metadata.Metadata{Type: "counter", Unit: "unit_4", Help: "help_4"}
m5 := metadata.Metadata{Type: "counter", Unit: "unit_5", Help: "help_5"}
app = db.Appender(ctx)
updateMetadata(t, app, s1, m1)
updateMetadata(t, app, s4, m4)
updateMetadata(t, app, s2, m5)
require.NoError(t, app.Commit())
series1 = db.head.series.getByHash(s1.Hash(), s1)
series2 = db.head.series.getByHash(s2.Hash(), s2)
series3 = db.head.series.getByHash(s3.Hash(), s3)
series4 = db.head.series.getByHash(s4.Hash(), s4)
require.Equal(t, *series1.meta, m1)
require.Equal(t, *series2.meta, m5)
require.Equal(t, *series3.meta, m3)
require.Equal(t, *series4.meta, m4)
Introduce TSDB changes for appending metadata to the WAL (#10972) * Append metadata to the WAL Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove extra whitespace; Reword some docstrings and comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use RLock() for hasNewMetadata check Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use single byte for metric type in RefMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update proposed WAL format for single-byte type metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implementa MetadataAppender interface for the Agent Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of review comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Amend description of metadata in wal.md Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Correct key used to retrieve metadata from cache When we're setting metadata entries in the scrapeCace, we're using the p.Help(), p.Unit(), p.Type() helpers, which retrieve the series name and use it as the cache key. When checking for cache entries though, we used p.Series() as the key, which included the metric name _with_ its labels. That meant that we were never actually hitting the cache. We're fixing this by utiling the __name__ internal label for correctly getting the cache entries after they've been set by setHelp(), setType() or setUnit(). Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Put feature behind a feature flag Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix AppendMetadata docstring Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reorder WAL format document Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Change error message of AppendMetadata; Fix access of s.meta in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse temporary buffer in Metadata encoder Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Only keep latest metadata for each refID during checkpointing Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test that's referencing decoding metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Avoid creating metadata block if no new metadata are present Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for corrupt metadata block and relevant record type Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix CR comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Extract logic about changing metadata in an anonymous function Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implement new proposed WAL format and amend relevant tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use 'const' for metadata field names Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Apply metadata to head memSeries in Commit, not in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add docstring and rename extracted helper in scrape.go Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for tsdb-related cases Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol1 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol2 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix Windows test by closing WAL reader files Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use switch instead of two if statements in metadata decoding Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments around TestMetadata* tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add code for replaying WAL; test correctness of in-memory data after a replay Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove scrape-loop related code from PR Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify tests by sorting slices before comparison Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test to use separate transactions Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Empty out buffer and record slices after encoding latest metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linting issue Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update calculation for DroppedMetadata metric Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Rename MetadataAppender interface and AppendMetadata method to MetadataUpdater/UpdateMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse buffer when encoding latest metadata for each series Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments; Check all returned error values using two helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify use of helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Satisfy linter Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com>
2022-07-19 01:58:52 -07:00
require.NoError(t, db.Close())
// Reopen the DB, replaying the WAL. The Head must have been replayed
// correctly in memory.
reopenDB, err := Open(db.Dir(), nil, nil, nil, nil)
require.NoError(t, err)
t.Cleanup(func() {
require.NoError(t, reopenDB.Close())
})
_, err = reopenDB.head.wal.Size()
require.NoError(t, err)
require.Equal(t, *reopenDB.head.series.getByHash(s1.Hash(), s1).meta, m1)
require.Equal(t, *reopenDB.head.series.getByHash(s2.Hash(), s2).meta, m5)
require.Equal(t, *reopenDB.head.series.getByHash(s3.Hash(), s3).meta, m3)
require.Equal(t, *reopenDB.head.series.getByHash(s4.Hash(), s4).meta, m4)
Introduce TSDB changes for appending metadata to the WAL (#10972) * Append metadata to the WAL Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove extra whitespace; Reword some docstrings and comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use RLock() for hasNewMetadata check Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use single byte for metric type in RefMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update proposed WAL format for single-byte type metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implementa MetadataAppender interface for the Agent Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of review comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Amend description of metadata in wal.md Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Correct key used to retrieve metadata from cache When we're setting metadata entries in the scrapeCace, we're using the p.Help(), p.Unit(), p.Type() helpers, which retrieve the series name and use it as the cache key. When checking for cache entries though, we used p.Series() as the key, which included the metric name _with_ its labels. That meant that we were never actually hitting the cache. We're fixing this by utiling the __name__ internal label for correctly getting the cache entries after they've been set by setHelp(), setType() or setUnit(). Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Put feature behind a feature flag Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix AppendMetadata docstring Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reorder WAL format document Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Change error message of AppendMetadata; Fix access of s.meta in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse temporary buffer in Metadata encoder Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Only keep latest metadata for each refID during checkpointing Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test that's referencing decoding metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Avoid creating metadata block if no new metadata are present Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for corrupt metadata block and relevant record type Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix CR comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Extract logic about changing metadata in an anonymous function Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Implement new proposed WAL format and amend relevant tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use 'const' for metadata field names Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Apply metadata to head memSeries in Commit, not in AppendMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add docstring and rename extracted helper in scrape.go Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add tests for tsdb-related cases Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol1 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linter issues vol2 Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix Windows test by closing WAL reader files Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Use switch instead of two if statements in metadata decoding Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments around TestMetadata* tests Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Add code for replaying WAL; test correctness of in-memory data after a replay Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Remove scrape-loop related code from PR Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Address first round of comments Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify tests by sorting slices before comparison Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix test to use separate transactions Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Empty out buffer and record slices after encoding latest metadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix linting issue Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Update calculation for DroppedMetadata metric Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Rename MetadataAppender interface and AppendMetadata method to MetadataUpdater/UpdateMetadata Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Reuse buffer when encoding latest metadata for each series Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Fix review comments; Check all returned error values using two helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Simplify use of helpers Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com> * Satisfy linter Signed-off-by: Paschalis Tsilias <paschalist0@gmail.com>
2022-07-19 01:58:52 -07:00
}