influxdb/tsdb/index_test.go

701 lines
21 KiB
Go

package tsdb_test
import (
"compress/gzip"
"fmt"
"io"
"os"
"path/filepath"
"reflect"
"sync"
"testing"
"github.com/influxdata/influxdb/v2/influxql/query"
"github.com/influxdata/influxdb/v2/internal"
"github.com/influxdata/influxdb/v2/models"
"github.com/influxdata/influxdb/v2/pkg/slices"
"github.com/influxdata/influxdb/v2/tsdb"
"github.com/influxdata/influxdb/v2/tsdb/index/tsi1"
"github.com/influxdata/influxql"
"go.uber.org/zap/zaptest"
)
// Ensure iterator can merge multiple iterators together.
func TestMergeSeriesIDIterators(t *testing.T) {
itr := tsdb.MergeSeriesIDIterators(
tsdb.NewSeriesIDSliceIterator([]uint64{1, 2, 3}),
tsdb.NewSeriesIDSliceIterator(nil),
nil,
tsdb.NewSeriesIDSliceIterator([]uint64{1, 2, 3, 4}),
)
if e, err := itr.Next(); err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(e, tsdb.SeriesIDElem{SeriesID: 1}) {
t.Fatalf("unexpected elem(0): %#v", e)
}
if e, err := itr.Next(); err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(e, tsdb.SeriesIDElem{SeriesID: 2}) {
t.Fatalf("unexpected elem(1): %#v", e)
}
if e, err := itr.Next(); err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(e, tsdb.SeriesIDElem{SeriesID: 3}) {
t.Fatalf("unexpected elem(2): %#v", e)
}
if e, err := itr.Next(); err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(e, tsdb.SeriesIDElem{SeriesID: 4}) {
t.Fatalf("unexpected elem(3): %#v", e)
}
if e, err := itr.Next(); err != nil {
t.Fatal(err)
} else if e.SeriesID != 0 {
t.Fatalf("expected nil elem: %#v", e)
}
}
func TestIndexSet_MeasurementNamesByExpr(t *testing.T) {
// Setup indexes
indexes := map[string]*Index{}
for _, name := range tsdb.RegisteredIndexes() {
idx := MustOpenNewIndex(t, name)
idx.AddSeries("cpu", map[string]string{"region": "east"})
idx.AddSeries("cpu", map[string]string{"region": "west", "secret": "foo"})
idx.AddSeries("disk", map[string]string{"secret": "foo"})
idx.AddSeries("mem", map[string]string{"region": "west"})
idx.AddSeries("gpu", map[string]string{"region": "east"})
idx.AddSeries("pci", map[string]string{"region": "east", "secret": "foo"})
indexes[name] = idx
defer idx.Close()
}
authorizer := &internal.AuthorizerMock{
AuthorizeSeriesReadFn: func(database string, measurement []byte, tags models.Tags) bool {
if tags.GetString("secret") != "" {
t.Logf("Rejecting series db=%s, m=%s, tags=%v", database, measurement, tags)
return false
}
return true
},
}
type example struct {
name string
expr influxql.Expr
expected [][]byte
}
// These examples should be run without any auth.
examples := []example{
{name: "all", expected: slices.StringsToBytes("cpu", "disk", "gpu", "mem", "pci")},
{name: "EQ", expr: influxql.MustParseExpr(`region = 'west'`), expected: slices.StringsToBytes("cpu", "mem")},
{name: "NEQ", expr: influxql.MustParseExpr(`region != 'west'`), expected: slices.StringsToBytes("gpu", "pci")},
{name: "EQREGEX", expr: influxql.MustParseExpr(`region =~ /.*st/`), expected: slices.StringsToBytes("cpu", "gpu", "mem", "pci")},
{name: "NEQREGEX", expr: influxql.MustParseExpr(`region !~ /.*est/`), expected: slices.StringsToBytes("gpu", "pci")},
}
// These examples should be run with the authorizer.
authExamples := []example{
{name: "all", expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "EQ", expr: influxql.MustParseExpr(`region = 'west'`), expected: slices.StringsToBytes("mem")},
{name: "NEQ", expr: influxql.MustParseExpr(`region != 'west'`), expected: slices.StringsToBytes("gpu")},
{name: "EQREGEX", expr: influxql.MustParseExpr(`region =~ /.*st/`), expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "NEQREGEX", expr: influxql.MustParseExpr(`region !~ /.*est/`), expected: slices.StringsToBytes("gpu")},
}
for _, idx := range tsdb.RegisteredIndexes() {
t.Run(idx, func(t *testing.T) {
t.Run("no authorization", func(t *testing.T) {
for _, example := range examples {
t.Run(example.name, func(t *testing.T) {
names, err := indexes[idx].IndexSet().MeasurementNamesByExpr(nil, example.expr)
if err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(names, example.expected) {
t.Fatalf("got names: %v, expected %v", slices.BytesToStrings(names), slices.BytesToStrings(example.expected))
}
})
}
})
t.Run("with authorization", func(t *testing.T) {
for _, example := range authExamples {
t.Run(example.name, func(t *testing.T) {
names, err := indexes[idx].IndexSet().MeasurementNamesByExpr(authorizer, example.expr)
if err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(names, example.expected) {
t.Fatalf("got names: %v, expected %v", slices.BytesToStrings(names), slices.BytesToStrings(example.expected))
}
})
}
})
})
}
}
func TestIndexSet_MeasurementNamesByPredicate(t *testing.T) {
// Setup indexes
indexes := map[string]*Index{}
for _, name := range tsdb.RegisteredIndexes() {
idx := MustOpenNewIndex(t, name)
idx.AddSeries("cpu", map[string]string{"region": "east"})
idx.AddSeries("cpu", map[string]string{"region": "west", "secret": "foo"})
idx.AddSeries("disk", map[string]string{"secret": "foo"})
idx.AddSeries("mem", map[string]string{"region": "west"})
idx.AddSeries("gpu", map[string]string{"region": "east"})
idx.AddSeries("pci", map[string]string{"region": "east", "secret": "foo"})
indexes[name] = idx
defer idx.Close()
}
authorizer := &internal.AuthorizerMock{
AuthorizeSeriesReadFn: func(database string, measurement []byte, tags models.Tags) bool {
if tags.GetString("secret") != "" {
t.Logf("Rejecting series db=%s, m=%s, tags=%v", database, measurement, tags)
return false
}
return true
},
}
type example struct {
name string
expr influxql.Expr
expected [][]byte
}
// These examples should be run without any auth.
examples := []example{
{name: "all", expected: slices.StringsToBytes("cpu", "disk", "gpu", "mem", "pci")},
{name: "EQ", expr: influxql.MustParseExpr(`region = 'west'`), expected: slices.StringsToBytes("cpu", "mem")},
{name: "NEQ", expr: influxql.MustParseExpr(`region != 'west'`), expected: slices.StringsToBytes("cpu", "disk", "gpu", "pci")},
{name: "EQREGEX", expr: influxql.MustParseExpr(`region =~ /.*st/`), expected: slices.StringsToBytes("cpu", "gpu", "mem", "pci")},
{name: "NEQREGEX", expr: influxql.MustParseExpr(`region !~ /.*est/`), expected: slices.StringsToBytes("cpu", "disk", "gpu", "pci")},
// None of the series have this tag so all should be selected.
{name: "EQ empty", expr: influxql.MustParseExpr(`host = ''`), expected: slices.StringsToBytes("cpu", "disk", "gpu", "mem", "pci")},
// Measurements that have this tag at all should be returned.
{name: "NEQ empty", expr: influxql.MustParseExpr(`region != ''`), expected: slices.StringsToBytes("cpu", "gpu", "mem", "pci")},
{name: "EQREGEX empty", expr: influxql.MustParseExpr(`host =~ /.*/`), expected: slices.StringsToBytes("cpu", "disk", "gpu", "mem", "pci")},
{name: "NEQ empty", expr: influxql.MustParseExpr(`region !~ /.*/`), expected: slices.StringsToBytes()},
}
// These examples should be run with the authorizer.
authExamples := []example{
{name: "all", expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "EQ", expr: influxql.MustParseExpr(`region = 'west'`), expected: slices.StringsToBytes("mem")},
{name: "NEQ", expr: influxql.MustParseExpr(`region != 'west'`), expected: slices.StringsToBytes("cpu", "gpu")},
{name: "EQREGEX", expr: influxql.MustParseExpr(`region =~ /.*st/`), expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "NEQREGEX", expr: influxql.MustParseExpr(`region !~ /.*est/`), expected: slices.StringsToBytes("cpu", "gpu")},
{name: "EQ empty", expr: influxql.MustParseExpr(`host = ''`), expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "NEQ empty", expr: influxql.MustParseExpr(`region != ''`), expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "EQREGEX empty", expr: influxql.MustParseExpr(`host =~ /.*/`), expected: slices.StringsToBytes("cpu", "gpu", "mem")},
{name: "NEQ empty", expr: influxql.MustParseExpr(`region !~ /.*/`), expected: slices.StringsToBytes()},
}
for _, idx := range tsdb.RegisteredIndexes() {
t.Run(idx, func(t *testing.T) {
t.Run("no authorization", func(t *testing.T) {
for _, example := range examples {
t.Run(example.name, func(t *testing.T) {
names, err := indexes[idx].IndexSet().MeasurementNamesByPredicate(nil, example.expr)
if err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(names, example.expected) {
t.Fatalf("got names: %v, expected %v", slices.BytesToStrings(names), slices.BytesToStrings(example.expected))
}
})
}
})
t.Run("with authorization", func(t *testing.T) {
for _, example := range authExamples {
t.Run(example.name, func(t *testing.T) {
names, err := indexes[idx].IndexSet().MeasurementNamesByPredicate(authorizer, example.expr)
if err != nil {
t.Fatal(err)
} else if !reflect.DeepEqual(names, example.expected) {
t.Fatalf("got names: %v, expected %v", slices.BytesToStrings(names), slices.BytesToStrings(example.expected))
}
})
}
})
})
}
}
func TestIndex_Sketches(t *testing.T) {
checkCardinalities := func(t *testing.T, index *Index, state string, series, tseries, measurements, tmeasurements int) {
t.Helper()
// Get sketches and check cardinality...
sketch, tsketch, err := index.SeriesSketches()
if err != nil {
t.Fatal(err)
}
// delta calculates a rough 10% delta. If i is small then a minimum value
// of 2 is used.
delta := func(i int) int {
v := i / 10
if v == 0 {
v = 2
}
return v
}
// series cardinality should be well within 10%.
if got, exp := int(sketch.Count()), series; got-exp < -delta(series) || got-exp > delta(series) {
t.Errorf("[%s] got series cardinality %d, expected ~%d", state, got, exp)
}
// check series tombstones
if got, exp := int(tsketch.Count()), tseries; got-exp < -delta(tseries) || got-exp > delta(tseries) {
t.Errorf("[%s] got series tombstone cardinality %d, expected ~%d", state, got, exp)
}
// Check measurement cardinality.
if sketch, tsketch, err = index.MeasurementsSketches(); err != nil {
t.Fatal(err)
}
if got, exp := int(sketch.Count()), measurements; got != exp { //got-exp < -delta(measurements) || got-exp > delta(measurements) {
t.Errorf("[%s] got measurement cardinality %d, expected ~%d", state, got, exp)
}
if got, exp := int(tsketch.Count()), tmeasurements; got != exp { //got-exp < -delta(tmeasurements) || got-exp > delta(tmeasurements) {
t.Errorf("[%s] got measurement tombstone cardinality %d, expected ~%d", state, got, exp)
}
}
test := func(t *testing.T, index string) error {
idx := MustNewIndex(t, index)
if index, ok := idx.Index.(*tsi1.Index); ok {
// Override the log file max size to force a log file compaction sooner.
// This way, we will test the sketches are correct when they have been
// compacted into IndexFiles, and also when they're loaded from
// IndexFiles after a re-open.
tsi1.WithMaximumLogFileSize(1 << 10)(index)
}
// Open the index
idx.MustOpen()
defer idx.Close()
series := genTestSeries(10, 5, 3)
// Add series to index.
for _, serie := range series {
if err := idx.AddSeries(serie.Measurement, serie.Tags.Map()); err != nil {
t.Fatal(err)
}
}
// Check cardinalities after adding series.
checkCardinalities(t, idx, "initial", 2430, 0, 10, 0)
// Re-open step only applies to the TSI index.
if _, ok := idx.Index.(*tsi1.Index); ok {
// Re-open the index.
if err := idx.Reopen(); err != nil {
panic(err)
}
// Check cardinalities after the reopen
checkCardinalities(t, idx, "initial|reopen", 2430, 0, 10, 0)
}
// Drop some series
if err := idx.DropMeasurement([]byte("measurement2")); err != nil {
return err
} else if err := idx.DropMeasurement([]byte("measurement5")); err != nil {
return err
}
// Check cardinalities after the delete
checkCardinalities(t, idx, "initial|reopen|delete", 2430, 486, 10, 2)
// Re-open step only applies to the TSI index.
if _, ok := idx.Index.(*tsi1.Index); ok {
// Re-open the index.
if err := idx.Reopen(); err != nil {
panic(err)
}
// Check cardinalities after the reopen
checkCardinalities(t, idx, "initial|reopen|delete|reopen", 2430, 486, 10, 2)
}
return nil
}
for _, index := range tsdb.RegisteredIndexes() {
t.Run(index, func(t *testing.T) {
if err := test(t, index); err != nil {
t.Fatal(err)
}
})
}
}
// Index wraps a series file and index.
type Index struct {
tsdb.Index
rootPath string
indexType string
sfile *tsdb.SeriesFile
}
type EngineOption func(opts *tsdb.EngineOptions)
// DisableTSICache allows the caller to disable the TSI bitset cache during a test.
var DisableTSICache = func() EngineOption {
return func(opts *tsdb.EngineOptions) {
opts.Config.SeriesIDSetCacheSize = 0
}
}
// MustNewIndex will initialize a new index using the provide type. It creates
// everything under the same root directory so it can be cleanly removed on Close.
//
// The index will not be opened.
func MustNewIndex(tb testing.TB, index string, eopts ...EngineOption) *Index {
tb.Helper()
opts := tsdb.NewEngineOptions()
opts.IndexVersion = index
for _, opt := range eopts {
opt(&opts)
}
rootPath := tb.TempDir()
seriesPath, err := os.MkdirTemp(rootPath, tsdb.SeriesFileDirectory)
if err != nil {
panic(err)
}
sfile := tsdb.NewSeriesFile(seriesPath)
if err := sfile.Open(); err != nil {
panic(err)
}
i, err := tsdb.NewIndex(0, "db0", filepath.Join(rootPath, "index"), tsdb.NewSeriesIDSet(), sfile, opts)
if err != nil {
panic(err)
}
i.WithLogger(zaptest.NewLogger(tb))
idx := &Index{
Index: i,
indexType: index,
rootPath: rootPath,
sfile: sfile,
}
return idx
}
// MustOpenNewIndex will initialize a new index using the provide type and opens
// it.
func MustOpenNewIndex(tb testing.TB, index string, opts ...EngineOption) *Index {
tb.Helper()
idx := MustNewIndex(tb, index, opts...)
idx.MustOpen()
return idx
}
// MustOpen opens the underlying index or panics.
func (i *Index) MustOpen() {
if err := i.Index.Open(); err != nil {
panic(err)
}
}
func (idx *Index) IndexSet() *tsdb.IndexSet {
return &tsdb.IndexSet{Indexes: []tsdb.Index{idx.Index}, SeriesFile: idx.sfile}
}
func (idx *Index) AddSeries(name string, tags map[string]string) error {
t := models.NewTags(tags)
key := fmt.Sprintf("%s,%s", name, t.HashKey())
return idx.CreateSeriesIfNotExists([]byte(key), []byte(name), t)
}
// Reopen closes and re-opens the underlying index, without removing any data.
func (i *Index) Reopen() error {
if err := i.Index.Close(); err != nil {
return err
}
if err := i.sfile.Close(); err != nil {
return err
}
i.sfile = tsdb.NewSeriesFile(i.sfile.Path())
if err := i.sfile.Open(); err != nil {
return err
}
opts := tsdb.NewEngineOptions()
opts.IndexVersion = i.indexType
idx, err := tsdb.NewIndex(0, "db0", filepath.Join(i.rootPath, "index"), tsdb.NewSeriesIDSet(), i.sfile, opts)
if err != nil {
return err
}
i.Index = idx
return i.Index.Open()
}
// Close closes the index cleanly and removes all on-disk data.
func (i *Index) Close() error {
if err := i.Index.Close(); err != nil {
return err
}
if err := i.sfile.Close(); err != nil {
return err
}
//return os.RemoveAll(i.rootPath)
return nil
}
// This benchmark compares the TagSets implementation across index types.
//
// In the case of the TSI index, TagSets has to merge results across all several
// index partitions.
//
// Typical results on an i7 laptop.
//
// BenchmarkIndexSet_TagSets/1M_series/tsi1-8 100 18995530 ns/op 5221180 B/op 20379 allocs/op
func BenchmarkIndexSet_TagSets(b *testing.B) {
// Read line-protocol and coerce into tsdb format.
keys := make([][]byte, 0, 1e6)
names := make([][]byte, 0, 1e6)
tags := make([]models.Tags, 0, 1e6)
// 1M series generated with:
// $inch -b 10000 -c 1 -t 10,10,10,10,10,10 -f 1 -m 5 -p 1
fd, err := os.Open("testdata/line-protocol-1M.txt.gz")
if err != nil {
b.Fatal(err)
}
gzr, err := gzip.NewReader(fd)
if err != nil {
fd.Close()
b.Fatal(err)
}
data, err := io.ReadAll(gzr)
if err != nil {
b.Fatal(err)
}
if err := fd.Close(); err != nil {
b.Fatal(err)
}
points, err := models.ParsePoints(data)
if err != nil {
b.Fatal(err)
}
for _, pt := range points {
keys = append(keys, pt.Key())
names = append(names, pt.Name())
tags = append(tags, pt.Tags())
}
// setup writes all of the above points to the index.
setup := func(idx *Index) {
batchSize := 10000
for j := 0; j < 1; j++ {
for i := 0; i < len(keys); i += batchSize {
k := keys[i : i+batchSize]
n := names[i : i+batchSize]
t := tags[i : i+batchSize]
if err := idx.CreateSeriesListIfNotExists(k, n, t); err != nil {
b.Fatal(err)
}
}
}
}
var errResult error
// This benchmark will merge eight bitsets each containing ~10,000 series IDs.
b.Run("1M series", func(b *testing.B) {
b.ReportAllocs()
for _, indexType := range tsdb.RegisteredIndexes() {
idx := MustOpenNewIndex(b, indexType)
setup(idx)
name := []byte("m4")
opt := query.IteratorOptions{Condition: influxql.MustParseExpr(`"tag5"::tag = 'value0'`)}
indexSet := tsdb.IndexSet{
SeriesFile: idx.sfile,
Indexes: []tsdb.Index{idx.Index},
} // For TSI implementation
ts := func() ([]*query.TagSet, error) {
return indexSet.TagSets(idx.sfile, name, opt)
}
b.Run(indexType, func(b *testing.B) {
for i := 0; i < b.N; i++ {
// Will call TagSets on the appropriate implementation.
_, errResult = ts()
if errResult != nil {
b.Fatal(err)
}
}
})
if err := idx.Close(); err != nil {
b.Fatal(err)
}
}
})
}
// This benchmark concurrently writes series to the index and fetches cached bitsets.
// The idea is to emphasize the performance difference when bitset caching is on and off.
//
// Typical results for an i7 laptop
//
// BenchmarkIndex_ConcurrentWriteQuery/tsi1/queries_100000/cache-8 1 1645048376 ns/op 2215402840 B/op 23048978 allocs/op
// BenchmarkIndex_ConcurrentWriteQuery/tsi1/queries_100000/no_cache-8 1 22242155616 ns/op 28277544136 B/op 79620463 allocs/op
func BenchmarkIndex_ConcurrentWriteQuery(b *testing.B) {
// Read line-protocol and coerce into tsdb format.
keys := make([][]byte, 0, 1e6)
names := make([][]byte, 0, 1e6)
tags := make([]models.Tags, 0, 1e6)
// 1M series generated with:
// $inch -b 10000 -c 1 -t 10,10,10,10,10,10 -f 1 -m 5 -p 1
fd, err := os.Open("testdata/line-protocol-1M.txt.gz")
if err != nil {
b.Fatal(err)
}
gzr, err := gzip.NewReader(fd)
if err != nil {
fd.Close()
b.Fatal(err)
}
data, err := io.ReadAll(gzr)
if err != nil {
b.Fatal(err)
}
if err := fd.Close(); err != nil {
b.Fatal(err)
}
points, err := models.ParsePoints(data)
if err != nil {
b.Fatal(err)
}
for _, pt := range points {
keys = append(keys, pt.Key())
names = append(names, pt.Name())
tags = append(tags, pt.Tags())
}
runBenchmark := func(b *testing.B, index string, queryN int, useTSICache bool) {
var idx *Index
if !useTSICache {
idx = MustOpenNewIndex(b, index, DisableTSICache())
} else {
idx = MustOpenNewIndex(b, index)
}
var wg sync.WaitGroup
begin := make(chan struct{})
// Run concurrent iterator...
runIter := func() {
keys := [][]string{
{"m0", "tag2", "value4"},
{"m1", "tag3", "value5"},
{"m2", "tag4", "value6"},
{"m3", "tag0", "value8"},
{"m4", "tag5", "value0"},
}
<-begin // Wait for writes to land
for i := 0; i < queryN/5; i++ {
for _, key := range keys {
itr, err := idx.TagValueSeriesIDIterator([]byte(key[0]), []byte(key[1]), []byte(key[2]))
if err != nil {
b.Fatal(err)
}
if itr == nil {
panic("should not happen")
}
if err := itr.Close(); err != nil {
b.Fatal(err)
}
}
}
}
batchSize := 10000
wg.Add(1)
go func() { defer wg.Done(); runIter() }()
var once sync.Once
for j := 0; j < b.N; j++ {
for i := 0; i < len(keys); i += batchSize {
k := keys[i : i+batchSize]
n := names[i : i+batchSize]
t := tags[i : i+batchSize]
if err := idx.CreateSeriesListIfNotExists(k, n, t); err != nil {
b.Fatal(err)
}
once.Do(func() { close(begin) })
}
// Wait for queries to finish
wg.Wait()
// Reset the index...
b.StopTimer()
if err := idx.Close(); err != nil {
b.Fatal(err)
}
// Re-open everything
idx = MustOpenNewIndex(b, index)
wg.Add(1)
begin = make(chan struct{})
once = sync.Once{}
go func() { defer wg.Done(); runIter() }()
b.StartTimer()
}
}
queries := []int{1e5}
for _, indexType := range tsdb.RegisteredIndexes() {
b.Run(indexType, func(b *testing.B) {
for _, queryN := range queries {
b.Run(fmt.Sprintf("queries %d", queryN), func(b *testing.B) {
b.Run("cache", func(b *testing.B) {
runBenchmark(b, indexType, queryN, true)
})
b.Run("no cache", func(b *testing.B) {
runBenchmark(b, indexType, queryN, false)
})
})
}
})
}
}