prometheus/tsdb/wal_test.go
Josh Soref 91d76c8023 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 15:54:09 +01:00

569 lines
15 KiB
Go

// 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.
// +build !windows
package tsdb
import (
"encoding/binary"
"io"
"io/ioutil"
"math/rand"
"os"
"path"
"path/filepath"
"testing"
"time"
"github.com/go-kit/kit/log"
"github.com/prometheus/prometheus/pkg/labels"
"github.com/prometheus/prometheus/tsdb/fileutil"
"github.com/prometheus/prometheus/tsdb/record"
"github.com/prometheus/prometheus/tsdb/tombstones"
"github.com/prometheus/prometheus/tsdb/wal"
"github.com/prometheus/prometheus/util/testutil"
)
func TestSegmentWAL_cut(t *testing.T) {
tmpdir, err := ioutil.TempDir("", "test_wal_cut")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(tmpdir))
}()
// This calls cut() implicitly the first time without a previous tail.
w, err := OpenSegmentWAL(tmpdir, nil, 0, nil)
testutil.Ok(t, err)
testutil.Ok(t, w.write(WALEntrySeries, 1, []byte("Hello World!!")))
testutil.Ok(t, w.cut())
// Cutting creates a new file.
testutil.Equals(t, 2, len(w.files))
testutil.Ok(t, w.write(WALEntrySeries, 1, []byte("Hello World!!")))
testutil.Ok(t, w.Close())
for _, of := range w.files {
f, err := os.Open(of.Name())
testutil.Ok(t, err)
// Verify header data.
metab := make([]byte, 8)
_, err = f.Read(metab)
testutil.Ok(t, err)
testutil.Equals(t, WALMagic, binary.BigEndian.Uint32(metab[:4]))
testutil.Equals(t, WALFormatDefault, metab[4])
// We cannot actually check for correct pre-allocation as it is
// optional per filesystem and handled transparently.
et, flag, b, err := newWALReader(nil, nil).entry(f)
testutil.Ok(t, err)
testutil.Equals(t, WALEntrySeries, et)
testutil.Equals(t, byte(walSeriesSimple), flag)
testutil.Equals(t, []byte("Hello World!!"), b)
}
}
func TestSegmentWAL_Truncate(t *testing.T) {
const (
numMetrics = 20000
batch = 100
)
series, err := labels.ReadLabels(filepath.Join("testdata", "20kseries.json"), numMetrics)
testutil.Ok(t, err)
dir, err := ioutil.TempDir("", "test_wal_log_truncate")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(dir))
}()
w, err := OpenSegmentWAL(dir, nil, 0, nil)
testutil.Ok(t, err)
w.segmentSize = 10000
for i := 0; i < numMetrics; i += batch {
var rs []record.RefSeries
for j, s := range series[i : i+batch] {
rs = append(rs, record.RefSeries{Labels: s, Ref: uint64(i+j) + 1})
}
err := w.LogSeries(rs)
testutil.Ok(t, err)
}
// We mark the 2nd half of the files with a min timestamp that should discard
// them from the selection of compactable files.
for i, f := range w.files[len(w.files)/2:] {
f.maxTime = int64(1000 + i)
}
// All series in those files must be preserved regarding of the provided postings list.
boundarySeries := w.files[len(w.files)/2].minSeries
// We truncate while keeping every 2nd series.
keep := map[uint64]struct{}{}
for i := 1; i <= numMetrics; i += 2 {
keep[uint64(i)] = struct{}{}
}
keepf := func(id uint64) bool {
_, ok := keep[id]
return ok
}
err = w.Truncate(1000, keepf)
testutil.Ok(t, err)
var expected []record.RefSeries
for i := 1; i <= numMetrics; i++ {
if i%2 == 1 || uint64(i) >= boundarySeries {
expected = append(expected, record.RefSeries{Ref: uint64(i), Labels: series[i-1]})
}
}
// Call Truncate once again to see whether we can read the written file without
// creating a new WAL.
err = w.Truncate(1000, keepf)
testutil.Ok(t, err)
testutil.Ok(t, w.Close())
// The same again with a new WAL.
w, err = OpenSegmentWAL(dir, nil, 0, nil)
testutil.Ok(t, err)
var readSeries []record.RefSeries
r := w.Reader()
testutil.Ok(t, r.Read(func(s []record.RefSeries) {
readSeries = append(readSeries, s...)
}, nil, nil))
testutil.Equals(t, expected, readSeries)
}
// Symmetrical test of reading and writing to the WAL via its main interface.
func TestSegmentWAL_Log_Restore(t *testing.T) {
const (
numMetrics = 50
iterations = 5
stepSize = 5
)
// Generate testing data. It does not make semantic sense but
// for the purpose of this test.
series, err := labels.ReadLabels(filepath.Join("testdata", "20kseries.json"), numMetrics)
testutil.Ok(t, err)
dir, err := ioutil.TempDir("", "test_wal_log_restore")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(dir))
}()
var (
recordedSeries [][]record.RefSeries
recordedSamples [][]record.RefSample
recordedDeletes [][]tombstones.Stone
)
var totalSamples int
// Open WAL a bunch of times, validate all previous data can be read,
// write more data to it, close it.
for k := 0; k < numMetrics; k += numMetrics / iterations {
w, err := OpenSegmentWAL(dir, nil, 0, nil)
testutil.Ok(t, err)
// Set smaller segment size so we can actually write several files.
w.segmentSize = 1000 * 1000
r := w.Reader()
var (
resultSeries [][]record.RefSeries
resultSamples [][]record.RefSample
resultDeletes [][]tombstones.Stone
)
serf := func(series []record.RefSeries) {
if len(series) > 0 {
clsets := make([]record.RefSeries, len(series))
copy(clsets, series)
resultSeries = append(resultSeries, clsets)
}
}
smplf := func(smpls []record.RefSample) {
if len(smpls) > 0 {
csmpls := make([]record.RefSample, len(smpls))
copy(csmpls, smpls)
resultSamples = append(resultSamples, csmpls)
}
}
delf := func(stones []tombstones.Stone) {
if len(stones) > 0 {
cst := make([]tombstones.Stone, len(stones))
copy(cst, stones)
resultDeletes = append(resultDeletes, cst)
}
}
testutil.Ok(t, r.Read(serf, smplf, delf))
testutil.Equals(t, recordedSamples, resultSamples)
testutil.Equals(t, recordedSeries, resultSeries)
testutil.Equals(t, recordedDeletes, resultDeletes)
series := series[k : k+(numMetrics/iterations)]
// Insert in batches and generate different amounts of samples for each.
for i := 0; i < len(series); i += stepSize {
var samples []record.RefSample
var stones []tombstones.Stone
for j := 0; j < i*10; j++ {
samples = append(samples, record.RefSample{
Ref: uint64(j % 10000),
T: int64(j * 2),
V: rand.Float64(),
})
}
for j := 0; j < i*20; j++ {
ts := rand.Int63()
stones = append(stones, tombstones.Stone{Ref: rand.Uint64(), Intervals: tombstones.Intervals{{Mint: ts, Maxt: ts + rand.Int63n(10000)}}})
}
lbls := series[i : i+stepSize]
series := make([]record.RefSeries, 0, len(series))
for j, l := range lbls {
series = append(series, record.RefSeries{
Ref: uint64(i + j),
Labels: l,
})
}
testutil.Ok(t, w.LogSeries(series))
testutil.Ok(t, w.LogSamples(samples))
testutil.Ok(t, w.LogDeletes(stones))
if len(lbls) > 0 {
recordedSeries = append(recordedSeries, series)
}
if len(samples) > 0 {
recordedSamples = append(recordedSamples, samples)
totalSamples += len(samples)
}
if len(stones) > 0 {
recordedDeletes = append(recordedDeletes, stones)
}
}
testutil.Ok(t, w.Close())
}
}
func TestWALRestoreCorrupted_invalidSegment(t *testing.T) {
dir, err := ioutil.TempDir("", "test_wal_log_restore")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(dir))
}()
wal, err := OpenSegmentWAL(dir, nil, 0, nil)
testutil.Ok(t, err)
_, err = wal.createSegmentFile(filepath.Join(dir, "000000"))
testutil.Ok(t, err)
f, err := wal.createSegmentFile(filepath.Join(dir, "000001"))
testutil.Ok(t, err)
f2, err := wal.createSegmentFile(filepath.Join(dir, "000002"))
testutil.Ok(t, err)
testutil.Ok(t, f2.Close())
// Make header of second segment invalid.
_, err = f.WriteAt([]byte{1, 2, 3, 4}, 0)
testutil.Ok(t, err)
testutil.Ok(t, f.Close())
testutil.Ok(t, wal.Close())
_, err = OpenSegmentWAL(dir, log.NewLogfmtLogger(os.Stderr), 0, nil)
testutil.Ok(t, err)
fns, err := fileutil.ReadDir(dir)
testutil.Ok(t, err)
testutil.Equals(t, []string{"000000"}, fns)
}
// Test reading from a WAL that has been corrupted through various means.
func TestWALRestoreCorrupted(t *testing.T) {
cases := []struct {
name string
f func(*testing.T, *SegmentWAL)
}{
{
name: "truncate_checksum",
f: func(t *testing.T, w *SegmentWAL) {
f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0666)
testutil.Ok(t, err)
defer f.Close()
off, err := f.Seek(0, io.SeekEnd)
testutil.Ok(t, err)
testutil.Ok(t, f.Truncate(off-1))
},
},
{
name: "truncate_body",
f: func(t *testing.T, w *SegmentWAL) {
f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0666)
testutil.Ok(t, err)
defer f.Close()
off, err := f.Seek(0, io.SeekEnd)
testutil.Ok(t, err)
testutil.Ok(t, f.Truncate(off-8))
},
},
{
name: "body_content",
f: func(t *testing.T, w *SegmentWAL) {
f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0666)
testutil.Ok(t, err)
defer f.Close()
off, err := f.Seek(0, io.SeekEnd)
testutil.Ok(t, err)
// Write junk before checksum starts.
_, err = f.WriteAt([]byte{1, 2, 3, 4}, off-8)
testutil.Ok(t, err)
},
},
{
name: "checksum",
f: func(t *testing.T, w *SegmentWAL) {
f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0666)
testutil.Ok(t, err)
defer f.Close()
off, err := f.Seek(0, io.SeekEnd)
testutil.Ok(t, err)
// Write junk into checksum
_, err = f.WriteAt([]byte{1, 2, 3, 4}, off-4)
testutil.Ok(t, err)
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
// Generate testing data. It does not make semantic sense but
// for the purpose of this test.
dir, err := ioutil.TempDir("", "test_corrupted")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(dir))
}()
w, err := OpenSegmentWAL(dir, nil, 0, nil)
testutil.Ok(t, err)
testutil.Ok(t, w.LogSamples([]record.RefSample{{T: 1, V: 2}}))
testutil.Ok(t, w.LogSamples([]record.RefSample{{T: 2, V: 3}}))
testutil.Ok(t, w.cut())
// Sleep 2 seconds to avoid error where cut and test "cases" function may write or
// truncate the file out of orders as "cases" are not synchronized with cut.
// Hopefully cut will complete by 2 seconds.
time.Sleep(2 * time.Second)
testutil.Ok(t, w.LogSamples([]record.RefSample{{T: 3, V: 4}}))
testutil.Ok(t, w.LogSamples([]record.RefSample{{T: 5, V: 6}}))
testutil.Ok(t, w.Close())
// cut() truncates and fsyncs the first segment async. If it happens after
// the corruption we apply below, the corruption will be overwritten again.
// Fire and forget a sync to avoid flakiness.
w.files[0].Sync()
// Corrupt the second entry in the first file.
// After re-opening we must be able to read the first entry
// and the rest, including the second file, must be truncated for clean further
// writes.
c.f(t, w)
logger := log.NewLogfmtLogger(os.Stderr)
w2, err := OpenSegmentWAL(dir, logger, 0, nil)
testutil.Ok(t, err)
r := w2.Reader()
serf := func(l []record.RefSeries) {
testutil.Equals(t, 0, len(l))
}
// Weird hack to check order of reads.
i := 0
samplef := func(s []record.RefSample) {
if i == 0 {
testutil.Equals(t, []record.RefSample{{T: 1, V: 2}}, s)
i++
} else {
testutil.Equals(t, []record.RefSample{{T: 99, V: 100}}, s)
}
}
testutil.Ok(t, r.Read(serf, samplef, nil))
testutil.Ok(t, w2.LogSamples([]record.RefSample{{T: 99, V: 100}}))
testutil.Ok(t, w2.Close())
// We should see the first valid entry and the new one, everything after
// is truncated.
w3, err := OpenSegmentWAL(dir, logger, 0, nil)
testutil.Ok(t, err)
r = w3.Reader()
i = 0
testutil.Ok(t, r.Read(serf, samplef, nil))
})
}
}
func TestMigrateWAL_Empty(t *testing.T) {
// The migration procedure must properly deal with a zero-length segment,
// which is valid in the new format.
dir, err := ioutil.TempDir("", "walmigrate")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(dir))
}()
wdir := path.Join(dir, "wal")
// Initialize empty WAL.
w, err := wal.New(nil, nil, wdir, false)
testutil.Ok(t, err)
testutil.Ok(t, w.Close())
testutil.Ok(t, MigrateWAL(nil, wdir))
}
func TestMigrateWAL_Fuzz(t *testing.T) {
dir, err := ioutil.TempDir("", "walmigrate")
testutil.Ok(t, err)
defer func() {
testutil.Ok(t, os.RemoveAll(dir))
}()
wdir := path.Join(dir, "wal")
// Should pass if no WAL exists yet.
testutil.Ok(t, MigrateWAL(nil, wdir))
oldWAL, err := OpenSegmentWAL(wdir, nil, time.Minute, nil)
testutil.Ok(t, err)
// Write some data.
testutil.Ok(t, oldWAL.LogSeries([]record.RefSeries{
{Ref: 100, Labels: labels.FromStrings("abc", "def", "123", "456")},
{Ref: 1, Labels: labels.FromStrings("abc", "def2", "1234", "4567")},
}))
testutil.Ok(t, oldWAL.LogSamples([]record.RefSample{
{Ref: 1, T: 100, V: 200},
{Ref: 2, T: 300, V: 400},
}))
testutil.Ok(t, oldWAL.LogSeries([]record.RefSeries{
{Ref: 200, Labels: labels.FromStrings("xyz", "def", "foo", "bar")},
}))
testutil.Ok(t, oldWAL.LogSamples([]record.RefSample{
{Ref: 3, T: 100, V: 200},
{Ref: 4, T: 300, V: 400},
}))
testutil.Ok(t, oldWAL.LogDeletes([]tombstones.Stone{
{Ref: 1, Intervals: []tombstones.Interval{{Mint: 100, Maxt: 200}}},
}))
testutil.Ok(t, oldWAL.Close())
// Perform migration.
testutil.Ok(t, MigrateWAL(nil, wdir))
w, err := wal.New(nil, nil, wdir, false)
testutil.Ok(t, err)
// We can properly write some new data after migration.
var enc record.Encoder
testutil.Ok(t, w.Log(enc.Samples([]record.RefSample{
{Ref: 500, T: 1, V: 1},
}, nil)))
testutil.Ok(t, w.Close())
// Read back all data.
sr, err := wal.NewSegmentsReader(wdir)
testutil.Ok(t, err)
r := wal.NewReader(sr)
var res []interface{}
var dec record.Decoder
for r.Next() {
rec := r.Record()
switch dec.Type(rec) {
case record.Series:
s, err := dec.Series(rec, nil)
testutil.Ok(t, err)
res = append(res, s)
case record.Samples:
s, err := dec.Samples(rec, nil)
testutil.Ok(t, err)
res = append(res, s)
case record.Tombstones:
s, err := dec.Tombstones(rec, nil)
testutil.Ok(t, err)
res = append(res, s)
default:
t.Fatalf("unknown record type %d", dec.Type(rec))
}
}
testutil.Ok(t, r.Err())
testutil.Equals(t, []interface{}{
[]record.RefSeries{
{Ref: 100, Labels: labels.FromStrings("abc", "def", "123", "456")},
{Ref: 1, Labels: labels.FromStrings("abc", "def2", "1234", "4567")},
},
[]record.RefSample{{Ref: 1, T: 100, V: 200}, {Ref: 2, T: 300, V: 400}},
[]record.RefSeries{
{Ref: 200, Labels: labels.FromStrings("xyz", "def", "foo", "bar")},
},
[]record.RefSample{{Ref: 3, T: 100, V: 200}, {Ref: 4, T: 300, V: 400}},
[]tombstones.Stone{{Ref: 1, Intervals: []tombstones.Interval{{Mint: 100, Maxt: 200}}}},
[]record.RefSample{{Ref: 500, T: 1, V: 1}},
}, res)
// Migrating an already migrated WAL shouldn't do anything.
testutil.Ok(t, MigrateWAL(nil, wdir))
}