lib/storage: sort tag filters by the number of loops they need for the execution

This metric should work better than the filter execution duration, since it cannot be distorted
by concurrently running queries.
This commit is contained in:
Aliaksandr Valialkin 2021-02-18 12:47:36 +02:00
parent f95dd67a22
commit 711f8a5b8d
3 changed files with 85 additions and 89 deletions

View File

@ -6,7 +6,6 @@ import (
"errors"
"fmt"
"io"
"math"
"path/filepath"
"sort"
"sync"
@ -2047,7 +2046,7 @@ func (is *indexSearch) getTagFilterWithMinMetricIDsCount(tfs *TagFilters, maxMet
continue
}
metricIDs, err := is.getMetricIDsForTagFilter(tf, nil, maxMetrics)
metricIDs, _, err := is.getMetricIDsForTagFilter(tf, nil, maxMetrics)
if err != nil {
if err == errFallbackToMetricNameMatch {
// Skip tag filters requiring to scan for too many metrics.
@ -2369,34 +2368,36 @@ const (
var uselessTagFilterCacheValue = []byte("1")
func (is *indexSearch) getMetricIDsForTagFilter(tf *tagFilter, filter *uint64set.Set, maxMetrics int) (*uint64set.Set, error) {
func (is *indexSearch) getMetricIDsForTagFilter(tf *tagFilter, filter *uint64set.Set, maxMetrics int) (*uint64set.Set, uint64, error) {
if tf.isNegative {
logger.Panicf("BUG: isNegative must be false")
}
metricIDs := &uint64set.Set{}
if len(tf.orSuffixes) > 0 {
// Fast path for orSuffixes - seek for rows for each value from orSuffixes.
if err := is.updateMetricIDsForOrSuffixesNoFilter(tf, maxMetrics, metricIDs); err != nil {
loopsCount, err := is.updateMetricIDsForOrSuffixesNoFilter(tf, maxMetrics, metricIDs)
if err != nil {
if err == errFallbackToMetricNameMatch {
return nil, err
return nil, loopsCount, err
}
return nil, fmt.Errorf("error when searching for metricIDs for tagFilter in fast path: %w; tagFilter=%s", err, tf)
return nil, loopsCount, fmt.Errorf("error when searching for metricIDs for tagFilter in fast path: %w; tagFilter=%s", err, tf)
}
return metricIDs, nil
return metricIDs, loopsCount, nil
}
// Slow path - scan for all the rows with the given prefix.
maxLoops := maxMetrics * maxIndexScanSlowLoopsPerMetric
if err := is.getMetricIDsForTagFilterSlow(tf, filter, maxLoops, metricIDs.Add); err != nil {
maxLoopsCount := uint64(maxMetrics) * maxIndexScanSlowLoopsPerMetric
loopsCount, err := is.getMetricIDsForTagFilterSlow(tf, filter, maxLoopsCount, metricIDs.Add)
if err != nil {
if err == errFallbackToMetricNameMatch {
return nil, err
return nil, loopsCount, err
}
return nil, fmt.Errorf("error when searching for metricIDs for tagFilter in slow path: %w; tagFilter=%s", err, tf)
return nil, loopsCount, fmt.Errorf("error when searching for metricIDs for tagFilter in slow path: %w; tagFilter=%s", err, tf)
}
return metricIDs, nil
return metricIDs, loopsCount, nil
}
func (is *indexSearch) getMetricIDsForTagFilterSlow(tf *tagFilter, filter *uint64set.Set, maxLoops int, f func(metricID uint64)) error {
func (is *indexSearch) getMetricIDsForTagFilterSlow(tf *tagFilter, filter *uint64set.Set, maxLoopsCount uint64, f func(metricID uint64)) (uint64, error) {
if len(tf.orSuffixes) > 0 {
logger.Panicf("BUG: the getMetricIDsForTagFilterSlow must be called only for empty tf.orSuffixes; got %s", tf.orSuffixes)
}
@ -2408,40 +2409,40 @@ func (is *indexSearch) getMetricIDsForTagFilterSlow(tf *tagFilter, filter *uint6
mp.Reset()
var prevMatchingSuffix []byte
var prevMatch bool
loops := 0
var loopsCount uint64
loopsPaceLimiter := 0
prefix := tf.prefix
ts.Seek(prefix)
for ts.NextItem() {
if loopsPaceLimiter&paceLimiterMediumIterationsMask == 0 {
if err := checkSearchDeadlineAndPace(is.deadline); err != nil {
return err
return loopsCount, err
}
}
loopsPaceLimiter++
item := ts.Item
if !bytes.HasPrefix(item, prefix) {
return nil
return loopsCount, nil
}
tail := item[len(prefix):]
n := bytes.IndexByte(tail, tagSeparatorChar)
if n < 0 {
return fmt.Errorf("invalid tag->metricIDs line %q: cannot find tagSeparatorChar=%d", item, tagSeparatorChar)
return loopsCount, fmt.Errorf("invalid tag->metricIDs line %q: cannot find tagSeparatorChar=%d", item, tagSeparatorChar)
}
suffix := tail[:n+1]
tail = tail[n+1:]
if err := mp.InitOnlyTail(item, tail); err != nil {
return err
return loopsCount, err
}
mp.ParseMetricIDs()
loopsCount += uint64(mp.MetricIDsLen())
if loopsCount > maxLoopsCount {
return loopsCount, errFallbackToMetricNameMatch
}
if prevMatch && string(suffix) == string(prevMatchingSuffix) {
// Fast path: the same tag value found.
// There is no need in checking it again with potentially
// slow tf.matchSuffix, which may call regexp.
loops += mp.MetricIDsLen()
if loops > maxLoops {
return errFallbackToMetricNameMatch
}
for _, metricID := range mp.MetricIDs {
if filter != nil && !filter.Has(metricID) {
continue
@ -2455,11 +2456,11 @@ func (is *indexSearch) getMetricIDsForTagFilterSlow(tf *tagFilter, filter *uint6
// since the current row has no matching metricIDs.
continue
}
// Slow path: need tf.matchSuffix call.
ok, err := tf.matchSuffix(suffix)
loopsCount += reMatchCost
if err != nil {
return fmt.Errorf("error when matching %s against suffix %q: %w", tf, suffix, err)
return loopsCount, fmt.Errorf("error when matching %s against suffix %q: %w", tf, suffix, err)
}
if !ok {
prevMatch = false
@ -2474,18 +2475,16 @@ func (is *indexSearch) getMetricIDsForTagFilterSlow(tf *tagFilter, filter *uint6
// The last char in kb.B must be tagSeparatorChar. Just increment it
// in order to jump to the next tag value.
if len(kb.B) == 0 || kb.B[len(kb.B)-1] != tagSeparatorChar || tagSeparatorChar >= 0xff {
return fmt.Errorf("data corruption: the last char in k=%X must be %X", kb.B, tagSeparatorChar)
return loopsCount, fmt.Errorf("data corruption: the last char in k=%X must be %X", kb.B, tagSeparatorChar)
}
kb.B[len(kb.B)-1]++
ts.Seek(kb.B)
// Assume that a seek cost is equivalent to 100 ordinary loops.
loopsCount += 100
continue
}
prevMatch = true
prevMatchingSuffix = append(prevMatchingSuffix[:0], suffix...)
loops += mp.MetricIDsLen()
if loops > maxLoops {
return errFallbackToMetricNameMatch
}
for _, metricID := range mp.MetricIDs {
if filter != nil && !filter.Has(metricID) {
continue
@ -2494,29 +2493,32 @@ func (is *indexSearch) getMetricIDsForTagFilterSlow(tf *tagFilter, filter *uint6
}
}
if err := ts.Error(); err != nil {
return fmt.Errorf("error when searching for tag filter prefix %q: %w", prefix, err)
return loopsCount, fmt.Errorf("error when searching for tag filter prefix %q: %w", prefix, err)
}
return nil
return loopsCount, nil
}
func (is *indexSearch) updateMetricIDsForOrSuffixesNoFilter(tf *tagFilter, maxMetrics int, metricIDs *uint64set.Set) error {
func (is *indexSearch) updateMetricIDsForOrSuffixesNoFilter(tf *tagFilter, maxMetrics int, metricIDs *uint64set.Set) (uint64, error) {
if tf.isNegative {
logger.Panicf("BUG: isNegative must be false")
}
kb := kbPool.Get()
defer kbPool.Put(kb)
var loopsCount uint64
for _, orSuffix := range tf.orSuffixes {
kb.B = append(kb.B[:0], tf.prefix...)
kb.B = append(kb.B, orSuffix...)
kb.B = append(kb.B, tagSeparatorChar)
if err := is.updateMetricIDsForOrSuffixNoFilter(kb.B, maxMetrics, metricIDs); err != nil {
return err
lc, err := is.updateMetricIDsForOrSuffixNoFilter(kb.B, maxMetrics, metricIDs)
if err != nil {
return loopsCount, err
}
loopsCount += lc
if metricIDs.Len() >= maxMetrics {
return nil
return loopsCount, nil
}
}
return nil
return loopsCount, nil
}
func (is *indexSearch) updateMetricIDsForOrSuffixesWithFilter(tf *tagFilter, metricIDs, filter *uint64set.Set) error {
@ -2534,39 +2536,39 @@ func (is *indexSearch) updateMetricIDsForOrSuffixesWithFilter(tf *tagFilter, met
return nil
}
func (is *indexSearch) updateMetricIDsForOrSuffixNoFilter(prefix []byte, maxMetrics int, metricIDs *uint64set.Set) error {
func (is *indexSearch) updateMetricIDsForOrSuffixNoFilter(prefix []byte, maxMetrics int, metricIDs *uint64set.Set) (uint64, error) {
ts := &is.ts
mp := &is.mp
mp.Reset()
maxLoops := maxMetrics * maxIndexScanLoopsPerMetric
loops := 0
maxLoopsCount := uint64(maxMetrics) * maxIndexScanLoopsPerMetric
var loopsCount uint64
loopsPaceLimiter := 0
ts.Seek(prefix)
for metricIDs.Len() < maxMetrics && ts.NextItem() {
if loopsPaceLimiter&paceLimiterFastIterationsMask == 0 {
if err := checkSearchDeadlineAndPace(is.deadline); err != nil {
return err
return loopsCount, err
}
}
loopsPaceLimiter++
item := ts.Item
if !bytes.HasPrefix(item, prefix) {
return nil
return loopsCount, nil
}
if err := mp.InitOnlyTail(item, item[len(prefix):]); err != nil {
return err
return loopsCount, err
}
loops += mp.MetricIDsLen()
if loops > maxLoops {
return errFallbackToMetricNameMatch
loopsCount += uint64(mp.MetricIDsLen())
if loopsCount > maxLoopsCount {
return loopsCount, errFallbackToMetricNameMatch
}
mp.ParseMetricIDs()
metricIDs.AddMulti(mp.MetricIDs)
}
if err := ts.Error(); err != nil {
return fmt.Errorf("error when searching for tag filter prefix %q: %w", prefix, err)
return loopsCount, fmt.Errorf("error when searching for tag filter prefix %q: %w", prefix, err)
}
return nil
return loopsCount, nil
}
func (is *indexSearch) updateMetricIDsForOrSuffixWithFilter(prefix []byte, metricIDs *uint64set.Set, sortedFilter []uint64, isNegative bool) error {
@ -2578,8 +2580,8 @@ func (is *indexSearch) updateMetricIDsForOrSuffixWithFilter(prefix []byte, metri
ts := &is.ts
mp := &is.mp
mp.Reset()
maxLoops := len(sortedFilter) * maxIndexScanLoopsPerMetric
loops := 0
maxLoopsCount := uint64(len(sortedFilter)) * maxIndexScanLoopsPerMetric
var loopsCount uint64
loopsPaceLimiter := 0
ts.Seek(prefix)
var sf []uint64
@ -2610,8 +2612,8 @@ func (is *indexSearch) updateMetricIDsForOrSuffixWithFilter(prefix []byte, metri
return nil
}
sf = sortedFilter
loops += mp.MetricIDsLen()
if loops > maxLoops {
loopsCount += uint64(mp.MetricIDsLen())
if loopsCount > maxLoopsCount {
return errFallbackToMetricNameMatch
}
mp.ParseMetricIDs()
@ -2781,31 +2783,33 @@ func (is *indexSearch) tryUpdatingMetricIDsForDateRange(metricIDs *uint64set.Set
}
func (is *indexSearch) getMetricIDsForDateAndFilters(date uint64, tfs *TagFilters, maxMetrics int) (*uint64set.Set, error) {
// Sort tfs by the duration from previous queries.
// Sort tfs by loopsCount needed for performing each filter.
// This stats is usually collected from the previous queries.
// This way we limit the amount of work below by applying fast filters at first.
type tagFilterWithWeight struct {
tf *tagFilter
durationSeconds float64
loopsCount uint64
lastQueryTimestamp uint64
}
tfws := make([]tagFilterWithWeight, len(tfs.tfs))
currentTime := fasttime.UnixTimestamp()
for i := range tfs.tfs {
tf := &tfs.tfs[i]
durationSeconds, lastQueryTimestamp := is.getDurationAndTimestampForDateFilter(date, tf)
if durationSeconds == 0 {
// Assume that unknown tag filters can take quite big amounts of time.
durationSeconds = 1.0
loopsCount, lastQueryTimestamp := is.getLoopsCountAndTimestampForDateFilter(date, tf)
if currentTime > lastQueryTimestamp+60*60 {
// Reset loopsCount to 0 every hour for collecting updated stats for the tf.
loopsCount = 0
}
tfws[i] = tagFilterWithWeight{
tf: tf,
durationSeconds: durationSeconds,
loopsCount: loopsCount,
lastQueryTimestamp: lastQueryTimestamp,
}
}
sort.Slice(tfws, func(i, j int) bool {
a, b := &tfws[i], &tfws[j]
if a.durationSeconds != b.durationSeconds {
return a.durationSeconds < b.durationSeconds
if a.loopsCount != b.loopsCount {
return a.loopsCount < b.loopsCount
}
return a.tf.Less(b.tf)
})
@ -2872,7 +2876,7 @@ func (is *indexSearch) getMetricIDsForDateAndFilters(date uint64, tfs *TagFilter
// Short circuit - there is no need in applying the remaining filters to an empty set.
break
}
if float64(metricIDsLen)/metricNameMatchesPerSecond < tfw.durationSeconds {
if uint64(metricIDsLen)*maxIndexScanLoopsPerMetric < tfw.loopsCount {
// It should be faster performing metricName match on the remaining filters
// instead of scanning big number of entries in the inverted index for these filters.
tfsPostponed = append(tfsPostponed, tf)
@ -2908,12 +2912,6 @@ func (is *indexSearch) getMetricIDsForDateAndFilters(date uint64, tfs *TagFilter
return metricIDs, nil
}
// The estimated number of per-second loops inside updateMetricIDsByMetricNameMatch
//
// This value is used for determining when matching by metric name must be perfromed instead of matching
// by the remaining tag filters.
const metricNameMatchesPerSecond = 50000
func (is *indexSearch) storeDateMetricID(date, metricID uint64) error {
ii := getIndexItems()
defer putIndexItems(ii)
@ -3073,8 +3071,7 @@ func (is *indexSearch) getMetricIDsForDateTagFilter(tf *tagFilter, lastQueryTime
tfNew := *tf
tfNew.isNegative = false // isNegative for the original tf is handled by the caller.
tfNew.prefix = kb.B
startTime := time.Now()
metricIDs, err := is.getMetricIDsForTagFilter(&tfNew, filter, maxMetrics)
metricIDs, loopsCount, err := is.getMetricIDsForTagFilter(&tfNew, filter, maxMetrics)
kbPool.Put(kb)
currentTimestamp := fasttime.UnixTimestamp()
if currentTimestamp > lastQueryTimestamp+5 {
@ -3082,23 +3079,22 @@ func (is *indexSearch) getMetricIDsForDateTagFilter(tf *tagFilter, lastQueryTime
// Do not update it too frequently.
return metricIDs, err
}
// Store the duration for tag filter execution in the cache in order to sort tag filters
// Store the loopsCount for tag filter in the cache in order to sort tag filters
// in ascending durations on the next search.
durationSeconds := time.Since(startTime).Seconds()
if err != nil && durationSeconds < 10 {
// Set high duration for failing filter, so it is moved to the end of filter list.
durationSeconds = 10
if err != nil {
// Set high loopsCount for failing filter, so it is moved to the end of filter list.
loopsCount = 1 << 30
}
if metricIDs.Len() >= maxMetrics {
// Increase the duration for tag filter matching too many metrics,
// So next time it will be applied after filters matching lower number of metrics.
durationSeconds *= 2
// Increase loopsCount for tag filter matching too many metrics,
// So next time it is moved to the end of filter list.
loopsCount *= 2
}
is.storeDurationAndTimestampForDateFilter(date, tf, durationSeconds, currentTimestamp)
is.storeLoopsCountAndTimestampForDateFilter(date, tf, loopsCount, currentTimestamp)
return metricIDs, err
}
func (is *indexSearch) getDurationAndTimestampForDateFilter(date uint64, tf *tagFilter) (float64, uint64) {
func (is *indexSearch) getLoopsCountAndTimestampForDateFilter(date uint64, tf *tagFilter) (uint64, uint64) {
is.kb.B = appendDateTagFilterCacheKey(is.kb.B[:0], date, tf)
kb := kbPool.Get()
defer kbPool.Put(kb)
@ -3106,17 +3102,15 @@ func (is *indexSearch) getDurationAndTimestampForDateFilter(date uint64, tf *tag
if len(kb.B) != 16 {
return 0, 0
}
n := encoding.UnmarshalUint64(kb.B)
durationSeconds := math.Float64frombits(n)
loopsCount := encoding.UnmarshalUint64(kb.B)
timestamp := encoding.UnmarshalUint64(kb.B[8:])
return durationSeconds, timestamp
return loopsCount, timestamp
}
func (is *indexSearch) storeDurationAndTimestampForDateFilter(date uint64, tf *tagFilter, durationSeconds float64, timestamp uint64) {
func (is *indexSearch) storeLoopsCountAndTimestampForDateFilter(date uint64, tf *tagFilter, loopsCount, timestamp uint64) {
is.kb.B = appendDateTagFilterCacheKey(is.kb.B[:0], date, tf)
n := math.Float64bits(durationSeconds)
kb := kbPool.Get()
kb.B = encoding.MarshalUint64(kb.B[:0], n)
kb.B = encoding.MarshalUint64(kb.B[:0], loopsCount)
kb.B = encoding.MarshalUint64(kb.B, timestamp)
is.db.durationsPerDateTagFilterCache.Set(is.kb.B, kb.B)
kbPool.Put(kb)
@ -3243,8 +3237,8 @@ func (is *indexSearch) intersectMetricIDsWithTagFilterNocache(tf *tagFilter, fil
}
// Slow path - scan for all the rows with the given prefix.
maxLoops := filter.Len() * maxIndexScanSlowLoopsPerMetric
err := is.getMetricIDsForTagFilterSlow(tf, filter, maxLoops, func(metricID uint64) {
maxLoopsCount := uint64(filter.Len()) * maxIndexScanSlowLoopsPerMetric
_, err := is.getMetricIDsForTagFilterSlow(tf, filter, maxLoopsCount, func(metricID uint64) {
if tf.isNegative {
// filter must be equal to metricIDs
metricIDs.Del(metricID)

View File

@ -203,7 +203,7 @@ func (tb *table) MustClose() {
if n := atomic.LoadUint64(&ptw.refCount); n != 1 {
logger.Panicf("BUG: unexpected refCount=%d when closing the partition; probably there are pending searches", n)
}
ptw.pt.MustClose()
ptw.decRef()
}
// Release exclusive lock on the table.

View File

@ -221,7 +221,9 @@ type tagFilter struct {
value []byte
isNegative bool
isRegexp bool
matchCost uint64
// matchCost is a cost for matching a filter against a single string.
matchCost uint64
// Prefix always contains {nsPrefixTagToMetricIDs, key}.
// Additionally it contains: