prometheus/rules/rules_test.go

601 lines
19 KiB
Go
Raw Normal View History

// Copyright 2013 Prometheus Team
// 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.
package rules
import (
"fmt"
"path"
"strings"
"testing"
"time"
clientmodel "github.com/prometheus/client_golang/model"
"github.com/prometheus/prometheus/rules/ast"
"github.com/prometheus/prometheus/stats"
"github.com/prometheus/prometheus/storage/metric"
"github.com/prometheus/prometheus/utility/test"
)
var (
testEvalTime = testStartTime.Add(testSampleInterval * 10)
fixturesPath = "fixtures"
)
Use custom timestamp type for sample timestamps and related code. So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f
2013-10-28 06:35:02 -07:00
func annotateWithTime(lines []string, timestamp clientmodel.Timestamp) []string {
annotatedLines := []string{}
for _, line := range lines {
annotatedLines = append(annotatedLines, fmt.Sprintf(line, timestamp))
}
return annotatedLines
}
func vectorComparisonString(expected []string, actual []string) string {
separator := "\n--------------\n"
return fmt.Sprintf("Expected:%v%v%v\nActual:%v%v%v ",
separator,
strings.Join(expected, "\n"),
separator,
separator,
strings.Join(actual, "\n"),
separator)
}
type testTieredStorageCloser struct {
storage *metric.TieredStorage
directory test.Closer
}
func (t testTieredStorageCloser) Close() {
t.storage.Close()
t.directory.Close()
}
// This is copied from storage/metric/helpers_test.go, which is unfortunate but
// presently required to make things work.
func NewTestTieredStorage(t test.Tester) (storage *metric.TieredStorage, closer test.Closer) {
var directory test.TemporaryDirectory
directory = test.NewTemporaryDirectory("test_tiered_storage", t)
2013-05-14 07:17:49 -07:00
storage, err := metric.NewTieredStorage(2500, 1000, 5*time.Second, 0*time.Second, directory.Path())
if err != nil {
if storage != nil {
storage.Close()
}
directory.Close()
t.Fatalf("Error creating storage: %s", err)
}
if storage == nil {
directory.Close()
t.Fatalf("storage == nil")
}
started := make(chan bool)
go storage.Serve(started)
<-started
closer = &testTieredStorageCloser{
storage: storage,
directory: directory,
}
return
}
func newTestStorage(t test.Tester) (storage *metric.TieredStorage, closer test.Closer) {
storage, closer = NewTestTieredStorage(t)
if storage == nil {
t.Fatal("storage == nil")
}
storeMatrix(*storage, testMatrix)
return
}
func TestExpressions(t *testing.T) {
// Labels in expected output need to be alphabetically sorted.
var expressionTests = []struct {
expr string
output []string
shouldFail bool
checkOrder bool
fullRanges int
intervalRanges int
}{
{
expr: `SUM(http_requests)`,
output: []string{`http_requests => 3600 @[%v]`},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests{instance="0"}) BY(job)`,
output: []string{
`http_requests{job="api-server"} => 400 @[%v]`,
`http_requests{job="app-server"} => 1200 @[%v]`,
},
fullRanges: 0,
intervalRanges: 4,
}, {
expr: `SUM(http_requests{instance="0"}) BY(job) KEEPING_EXTRA`,
output: []string{
`http_requests{instance="0", job="api-server"} => 400 @[%v]`,
`http_requests{instance="0", job="app-server"} => 1200 @[%v]`,
},
fullRanges: 0,
intervalRanges: 4,
}, {
expr: `SUM(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 1000 @[%v]`,
`http_requests{job="app-server"} => 2600 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `
// Test comment.
SUM(http_requests) BY /* comments shouldn't
have any effect */ (job) // another comment`,
output: []string{
`http_requests{job="api-server"} => 1000 @[%v]`,
`http_requests{job="app-server"} => 2600 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `COUNT(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 4 @[%v]`,
`http_requests{job="app-server"} => 4 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job, group)`,
output: []string{
`http_requests{group="canary", job="api-server"} => 700 @[%v]`,
`http_requests{group="canary", job="app-server"} => 1500 @[%v]`,
`http_requests{group="production", job="api-server"} => 300 @[%v]`,
`http_requests{group="production", job="app-server"} => 1100 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `AVG(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 250 @[%v]`,
`http_requests{job="app-server"} => 650 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `MIN(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 100 @[%v]`,
`http_requests{job="app-server"} => 500 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `MAX(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 400 @[%v]`,
`http_requests{job="app-server"} => 800 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) - COUNT(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 996 @[%v]`,
`http_requests{job="app-server"} => 2596 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) - 2`,
output: []string{
`http_requests{job="api-server"} => 998 @[%v]`,
`http_requests{job="app-server"} => 2598 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) % 3`,
output: []string{
`http_requests{job="api-server"} => 1 @[%v]`,
`http_requests{job="app-server"} => 2 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) / 0`,
output: []string{
`http_requests{job="api-server"} => +Inf @[%v]`,
`http_requests{job="app-server"} => +Inf @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) > 1000`,
output: []string{
`http_requests{job="app-server"} => 2600 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) <= 1000`,
output: []string{
`http_requests{job="api-server"} => 1000 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) != 1000`,
output: []string{
`http_requests{job="app-server"} => 2600 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) == 1000`,
output: []string{
`http_requests{job="api-server"} => 1000 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `SUM(http_requests) BY (job) + SUM(http_requests) BY (job)`,
output: []string{
`http_requests{job="api-server"} => 2000 @[%v]`,
`http_requests{job="app-server"} => 5200 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `http_requests{job="api-server", group="canary"}`,
output: []string{
`http_requests{group="canary", instance="0", job="api-server"} => 300 @[%v]`,
`http_requests{group="canary", instance="1", job="api-server"} => 400 @[%v]`,
},
fullRanges: 0,
intervalRanges: 2,
}, {
expr: `http_requests{job="api-server", group="canary"} + delta(http_requests{job="api-server"}[5m], 1)`,
output: []string{
`http_requests{group="canary", instance="0", job="api-server"} => 330 @[%v]`,
`http_requests{group="canary", instance="1", job="api-server"} => 440 @[%v]`,
},
fullRanges: 4,
intervalRanges: 0,
}, {
expr: `delta(http_requests[25m], 1)`,
output: []string{
`http_requests{group="canary", instance="0", job="api-server"} => 150 @[%v]`,
`http_requests{group="canary", instance="0", job="app-server"} => 350 @[%v]`,
`http_requests{group="canary", instance="1", job="api-server"} => 200 @[%v]`,
`http_requests{group="canary", instance="1", job="app-server"} => 400 @[%v]`,
`http_requests{group="production", instance="0", job="api-server"} => 50 @[%v]`,
`http_requests{group="production", instance="0", job="app-server"} => 250 @[%v]`,
`http_requests{group="production", instance="1", job="api-server"} => 100 @[%v]`,
`http_requests{group="production", instance="1", job="app-server"} => 300 @[%v]`,
},
fullRanges: 8,
intervalRanges: 0,
}, {
expr: `sort(http_requests)`,
output: []string{
`http_requests{group="production", instance="0", job="api-server"} => 100 @[%v]`,
`http_requests{group="production", instance="1", job="api-server"} => 200 @[%v]`,
`http_requests{group="canary", instance="0", job="api-server"} => 300 @[%v]`,
`http_requests{group="canary", instance="1", job="api-server"} => 400 @[%v]`,
`http_requests{group="production", instance="0", job="app-server"} => 500 @[%v]`,
`http_requests{group="production", instance="1", job="app-server"} => 600 @[%v]`,
`http_requests{group="canary", instance="0", job="app-server"} => 700 @[%v]`,
`http_requests{group="canary", instance="1", job="app-server"} => 800 @[%v]`,
},
checkOrder: true,
fullRanges: 0,
intervalRanges: 8,
}, {
expr: `sort_desc(http_requests)`,
output: []string{
`http_requests{group="canary", instance="1", job="app-server"} => 800 @[%v]`,
`http_requests{group="canary", instance="0", job="app-server"} => 700 @[%v]`,
`http_requests{group="production", instance="1", job="app-server"} => 600 @[%v]`,
`http_requests{group="production", instance="0", job="app-server"} => 500 @[%v]`,
`http_requests{group="canary", instance="1", job="api-server"} => 400 @[%v]`,
`http_requests{group="canary", instance="0", job="api-server"} => 300 @[%v]`,
`http_requests{group="production", instance="1", job="api-server"} => 200 @[%v]`,
`http_requests{group="production", instance="0", job="api-server"} => 100 @[%v]`,
},
checkOrder: true,
fullRanges: 0,
intervalRanges: 8,
}, {
// Single-letter label names and values.
expr: `x{y="testvalue"}`,
output: []string{
`x{y="testvalue"} => 100 @[%v]`,
},
fullRanges: 0,
intervalRanges: 1,
}, {
// Lower-cased aggregation operators should work too.
expr: `sum(http_requests) by (job) + min(http_requests) by (job) + max(http_requests) by (job) + avg(http_requests) by (job)`,
output: []string{
`http_requests{job="app-server"} => 4550 @[%v]`,
`http_requests{job="api-server"} => 1750 @[%v]`,
},
fullRanges: 0,
intervalRanges: 8,
}, {
// Deltas should be adjusted for target interval vs. samples under target interval.
expr: `delta(http_requests{group="canary", instance="1", job="app-server"}[18m], 1)`,
output: []string{`http_requests{group="canary", instance="1", job="app-server"} => 288 @[%v]`},
fullRanges: 1,
intervalRanges: 0,
}, {
// Rates should transform per-interval deltas to per-second rates.
expr: `rate(http_requests{group="canary", instance="1", job="app-server"}[10m])`,
output: []string{`http_requests{group="canary", instance="1", job="app-server"} => 0.26666666666666666 @[%v]`},
fullRanges: 1,
intervalRanges: 0,
}, {
// Counter resets in middle of range are ignored by delta() if counter == 1.
expr: `delta(testcounter_reset_middle[50m], 1)`,
output: []string{`testcounter_reset_middle => 90 @[%v]`},
fullRanges: 1,
intervalRanges: 0,
}, {
// Counter resets in middle of range are not ignored by delta() if counter == 0.
expr: `delta(testcounter_reset_middle[50m], 0)`,
output: []string{`testcounter_reset_middle => 50 @[%v]`},
fullRanges: 1,
intervalRanges: 0,
}, {
// Counter resets at end of range are ignored by delta() if counter == 1.
expr: `delta(testcounter_reset_end[5m], 1)`,
output: []string{`testcounter_reset_end => 0 @[%v]`},
fullRanges: 1,
intervalRanges: 0,
}, {
// Counter resets at end of range are not ignored by delta() if counter == 0.
expr: `delta(testcounter_reset_end[5m], 0)`,
output: []string{`testcounter_reset_end => -90 @[%v]`},
fullRanges: 1,
intervalRanges: 0,
}, {
// Empty expressions shouldn"t parse.
expr: ``,
shouldFail: true,
}, {
// Subtracting a vector from a scalar is not supported.
expr: `1 - http_requests`,
shouldFail: true,
}, {
// Interval durations can"t be in quotes.
expr: `http_requests["1m"]`,
shouldFail: true,
},
}
tieredStorage, closer := newTestStorage(t)
defer closer.Close()
tieredStorage.Flush()
for i, exprTest := range expressionTests {
expectedLines := annotateWithTime(exprTest.output, testEvalTime)
testExpr, err := LoadExprFromString(exprTest.expr)
if err != nil {
if exprTest.shouldFail {
continue
}
t.Errorf("%d. Error during parsing: %v", i, err)
t.Errorf("%d. Expression: %v", i, exprTest.expr)
} else {
if exprTest.shouldFail {
t.Errorf("%d. Test should fail, but didn't", i)
}
failed := false
resultStr := ast.EvalToString(testExpr, testEvalTime, ast.TEXT, tieredStorage, stats.NewTimerGroup())
resultLines := strings.Split(resultStr, "\n")
if len(exprTest.output) != len(resultLines) {
t.Errorf("%d. Number of samples in expected and actual output don't match", i)
failed = true
}
if exprTest.checkOrder {
for j, expectedSample := range expectedLines {
if resultLines[j] != expectedSample {
t.Errorf("%d.%d. Expected sample '%v', got '%v'", i, j, resultLines[j], expectedSample)
failed = true
}
}
} else {
for j, expectedSample := range expectedLines {
found := false
for _, actualSample := range resultLines {
if actualSample == expectedSample {
found = true
}
}
if !found {
t.Errorf("%d.%d. Couldn't find expected sample in output: '%v'", i, j, expectedSample)
failed = true
}
}
}
2013-05-07 04:15:10 -07:00
analyzer := ast.NewQueryAnalyzer(tieredStorage)
analyzer.AnalyzeQueries(testExpr)
if exprTest.fullRanges != len(analyzer.FullRanges) {
t.Errorf("%d. Count of full ranges didn't match: %v vs %v", i, exprTest.fullRanges, len(analyzer.FullRanges))
failed = true
}
if exprTest.intervalRanges != len(analyzer.IntervalRanges) {
t.Errorf("%d. Count of interval ranges didn't match: %v vs %v", i, exprTest.intervalRanges, len(analyzer.IntervalRanges))
failed = true
}
if failed {
t.Errorf("%d. Expression: %v\n%v", i, exprTest.expr, vectorComparisonString(expectedLines, resultLines))
}
}
}
}
var ruleTests = []struct {
inputFile string
shouldFail bool
errContains string
numRecordingRules int
numAlertingRules int
}{
{
inputFile: "empty.rules",
numRecordingRules: 0,
numAlertingRules: 0,
}, {
inputFile: "mixed.rules",
numRecordingRules: 2,
numAlertingRules: 2,
},
{
inputFile: "syntax_error.rules",
shouldFail: true,
errContains: "Error parsing rules at line 5",
},
{
inputFile: "non_vector.rules",
shouldFail: true,
errContains: "does not evaluate to vector type",
},
}
func TestRules(t *testing.T) {
for i, ruleTest := range ruleTests {
testRules, err := LoadRulesFromFile(path.Join(fixturesPath, ruleTest.inputFile))
if err != nil {
if !ruleTest.shouldFail {
t.Fatalf("%d. Error parsing rules file %v: %v", i, ruleTest.inputFile, err)
} else {
if !strings.Contains(err.Error(), ruleTest.errContains) {
t.Fatalf("%d. Expected error containing '%v', got: %v", i, ruleTest.errContains, err)
}
}
} else {
numRecordingRules := 0
numAlertingRules := 0
for j, rule := range testRules {
switch rule.(type) {
case *RecordingRule:
numRecordingRules++
case *AlertingRule:
numAlertingRules++
default:
t.Fatalf("%d.%d. Unknown rule type!", i, j)
}
}
if numRecordingRules != ruleTest.numRecordingRules {
t.Fatalf("%d. Expected %d recording rules, got %d", i, ruleTest.numRecordingRules, numRecordingRules)
}
if numAlertingRules != ruleTest.numAlertingRules {
t.Fatalf("%d. Expected %d alerting rules, got %d", i, ruleTest.numAlertingRules, numAlertingRules)
}
// TODO(julius): add more complex checks on the parsed rules here.
}
}
}
func TestAlertingRule(t *testing.T) {
// Labels in expected output need to be alphabetically sorted.
var evalOutputs = [][]string{
{
`ALERTS{alertname="HttpRequestRateLow", alertstate="pending", group="canary", instance="0", job="app-server", severity="critical"} => 1 @[%v]`,
`ALERTS{alertname="HttpRequestRateLow", alertstate="pending", group="canary", instance="1", job="app-server", severity="critical"} => 1 @[%v]`,
},
{
`ALERTS{alertname="HttpRequestRateLow", alertstate="pending", group="canary", instance="0", job="app-server", severity="critical"} => 0 @[%v]`,
`ALERTS{alertname="HttpRequestRateLow", alertstate="firing", group="canary", instance="0", job="app-server", severity="critical"} => 1 @[%v]`,
`ALERTS{alertname="HttpRequestRateLow", alertstate="pending", group="canary", instance="1", job="app-server", severity="critical"} => 0 @[%v]`,
`ALERTS{alertname="HttpRequestRateLow", alertstate="firing", group="canary", instance="1", job="app-server", severity="critical"} => 1 @[%v]`,
},
{
`ALERTS{alertname="HttpRequestRateLow", alertstate="firing", group="canary", instance="1", job="app-server", severity="critical"} => 0 @[%v]`,
`ALERTS{alertname="HttpRequestRateLow", alertstate="firing", group="canary", instance="0", job="app-server", severity="critical"} => 0 @[%v]`,
},
{
/* empty */
},
{
/* empty */
},
}
tieredStorage, closer := newTestStorage(t)
defer closer.Close()
tieredStorage.Flush()
alertExpr, err := LoadExprFromString(`http_requests{group="canary", job="app-server"} < 100`)
if err != nil {
t.Fatalf("Unable to parse alert expression: %s", err)
}
alertName := "HttpRequestRateLow"
alertLabels := clientmodel.LabelSet{
"severity": "critical",
}
rule := NewAlertingRule(alertName, alertExpr.(ast.VectorNode), time.Minute, alertLabels, "summary", "description")
for i, expected := range evalOutputs {
evalTime := testStartTime.Add(testSampleInterval * time.Duration(i))
2013-05-07 04:15:10 -07:00
actual, err := rule.Eval(evalTime, tieredStorage)
if err != nil {
t.Fatalf("Error during alerting rule evaluation: %s", err)
}
actualLines := strings.Split(actual.String(), "\n")
expectedLines := annotateWithTime(expected, evalTime)
if actualLines[0] == "" {
actualLines = []string{}
}
failed := false
if len(actualLines) != len(expectedLines) {
t.Errorf("%d. Number of samples in expected and actual output don't match (%d vs. %d)", i, len(expectedLines), len(actualLines))
failed = true
}
for j, expectedSample := range expectedLines {
found := false
for _, actualSample := range actualLines {
if actualSample == expectedSample {
found = true
}
}
if !found {
t.Errorf("%d.%d. Couldn't find expected sample in output: '%v'", i, j, expectedSample)
failed = true
}
}
if failed {
t.Fatalf("%d. Expected and actual outputs don't match:\n%v", i, vectorComparisonString(expectedLines, actualLines))
}
}
}