VictoriaMetrics/lib/storage/search.go
Aliaksandr Valialkin 5e4dfe50c6
lib/storage: subsitute searchTSIDs functions with more lightweight searchMetricIDs function
The searchTSIDs function was searching for metricIDs matching the the given tag filters
and then was locating the corresponding TSID entries for the found metricIDs.

The TSID entries aren't needed when searching for time series names (aka MetricName),
so this commit removes the uneeded TSID search from the implementation of /api/v1/series API.
This improves perfromance of /api/v1/series calls.

This commit also improves performance a bit for /api/v1/query and /api/v1/query_range calls,
since now these calls cache small metricIDs instead of big TSID entries
in the indexdb/tagFilters cache (now this cache is named indexdb/tagFiltersToMetricIDs)
without the need to compress the saved entries in order to save cache space.

This commit also removes concurrency limiter during searching for matching time series,
which was introduced in 8f16388428, since the concurrency
for all the read queries is already limited with -search.maxConcurrentRequests command-line flag.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/648
2022-10-23 12:23:47 +03:00

449 lines
11 KiB
Go

package storage
import (
"fmt"
"io"
"strings"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/querytracer"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/storagepacelimiter"
)
// BlockRef references a Block.
//
// BlockRef is valid only until the corresponding Search is valid,
// i.e. it becomes invalid after Search.MustClose is called.
type BlockRef struct {
p *part
bh blockHeader
}
func (br *BlockRef) reset() {
br.p = nil
br.bh = blockHeader{}
}
func (br *BlockRef) init(p *part, bh *blockHeader) {
br.p = p
br.bh = *bh
}
// Init initializes br from pr and data
func (br *BlockRef) Init(pr PartRef, data []byte) error {
br.p = pr.p
tail, err := br.bh.Unmarshal(data)
if err != nil {
return err
}
if len(tail) > 0 {
return fmt.Errorf("unexpected non-empty tail left after unmarshaling blockHeader; len(tail)=%d; tail=%q", len(tail), tail)
}
return nil
}
// Marshal marshals br to dst.
func (br *BlockRef) Marshal(dst []byte) []byte {
return br.bh.Marshal(dst)
}
// RowsCount returns the number of rows in br.
func (br *BlockRef) RowsCount() int {
return int(br.bh.RowsCount)
}
// PartRef returns PartRef from br.
func (br *BlockRef) PartRef() PartRef {
return PartRef{
p: br.p,
}
}
// PartRef is Part reference.
type PartRef struct {
p *part
}
// MustReadBlock reads block from br to dst.
func (br *BlockRef) MustReadBlock(dst *Block) {
dst.Reset()
dst.bh = br.bh
dst.timestampsData = bytesutil.ResizeNoCopyMayOverallocate(dst.timestampsData, int(br.bh.TimestampsBlockSize))
br.p.timestampsFile.MustReadAt(dst.timestampsData, int64(br.bh.TimestampsBlockOffset))
dst.valuesData = bytesutil.ResizeNoCopyMayOverallocate(dst.valuesData, int(br.bh.ValuesBlockSize))
br.p.valuesFile.MustReadAt(dst.valuesData, int64(br.bh.ValuesBlockOffset))
}
// MetricBlockRef contains reference to time series block for a single metric.
type MetricBlockRef struct {
// The metric name
MetricName []byte
// The block reference. Call BlockRef.MustReadBlock in order to obtain the block.
BlockRef *BlockRef
}
// Search is a search for time series.
type Search struct {
// MetricBlockRef is updated with each Search.NextMetricBlock call.
MetricBlockRef MetricBlockRef
// idb is used for MetricName lookup for the found data blocks.
idb *indexDB
ts tableSearch
// tr contains time range used in the serach.
tr TimeRange
// tfss contains tag filters used in the search.
tfss []*TagFilters
// deadline in unix timestamp seconds for the current search.
deadline uint64
err error
needClosing bool
loops int
prevMetricID uint64
}
func (s *Search) reset() {
s.MetricBlockRef.MetricName = s.MetricBlockRef.MetricName[:0]
s.MetricBlockRef.BlockRef = nil
s.idb = nil
s.ts.reset()
s.tr = TimeRange{}
s.tfss = nil
s.deadline = 0
s.err = nil
s.needClosing = false
s.loops = 0
s.prevMetricID = 0
}
// Init initializes s from the given storage, tfss and tr.
//
// MustClose must be called when the search is done.
//
// Init returns the upper bound on the number of found time series.
func (s *Search) Init(qt *querytracer.Tracer, storage *Storage, tfss []*TagFilters, tr TimeRange, maxMetrics int, deadline uint64) int {
qt = qt.NewChild("init series search: filters=%s, timeRange=%s", tfss, &tr)
defer qt.Done()
if s.needClosing {
logger.Panicf("BUG: missing MustClose call before the next call to Init")
}
s.reset()
s.idb = storage.idb()
s.tr = tr
s.tfss = tfss
s.deadline = deadline
s.needClosing = true
var tsids []TSID
metricIDs, err := s.idb.searchMetricIDs(qt, tfss, tr, maxMetrics, deadline)
if err == nil {
tsids, err = s.idb.getTSIDsFromMetricIDs(qt, metricIDs, deadline)
if err == nil {
err = storage.prefetchMetricNames(qt, metricIDs, deadline)
}
}
// It is ok to call Init on non-nil err.
// Init must be called before returning because it will fail
// on Seach.MustClose otherwise.
s.ts.Init(storage.tb, tsids, tr)
qt.Printf("search for parts with data for %d series", len(tsids))
if err != nil {
s.err = err
return 0
}
return len(tsids)
}
// MustClose closes the Search.
func (s *Search) MustClose() {
if !s.needClosing {
logger.Panicf("BUG: missing Init call before MustClose")
}
s.ts.MustClose()
s.reset()
}
// Error returns the last error from s.
func (s *Search) Error() error {
if s.err == io.EOF || s.err == nil {
return nil
}
return fmt.Errorf("error when searching for tagFilters=%s on the time range %s: %w", s.tfss, s.tr.String(), s.err)
}
// NextMetricBlock proceeds to the next MetricBlockRef.
func (s *Search) NextMetricBlock() bool {
if s.err != nil {
return false
}
for s.ts.NextBlock() {
if s.loops&paceLimiterSlowIterationsMask == 0 {
if err := checkSearchDeadlineAndPace(s.deadline); err != nil {
s.err = err
return false
}
}
s.loops++
tsid := &s.ts.BlockRef.bh.TSID
if tsid.MetricID != s.prevMetricID {
var err error
s.MetricBlockRef.MetricName, err = s.idb.searchMetricNameWithCache(s.MetricBlockRef.MetricName[:0], tsid.MetricID)
if err != nil {
if err == io.EOF {
// Skip missing metricName for tsid.MetricID.
// It should be automatically fixed. See indexDB.searchMetricNameWithCache for details.
continue
}
s.err = err
return false
}
s.prevMetricID = tsid.MetricID
}
s.MetricBlockRef.BlockRef = s.ts.BlockRef
return true
}
if err := s.ts.Error(); err != nil {
s.err = err
return false
}
s.err = io.EOF
return false
}
// SearchQuery is used for sending search queries from vmselect to vmstorage.
type SearchQuery struct {
// The time range for searching time series
MinTimestamp int64
MaxTimestamp int64
// Tag filters for the search query
TagFilterss [][]TagFilter
// The maximum number of time series the search query can return.
MaxMetrics int
}
// GetTimeRange returns time range for the given sq.
func (sq *SearchQuery) GetTimeRange() TimeRange {
return TimeRange{
MinTimestamp: sq.MinTimestamp,
MaxTimestamp: sq.MaxTimestamp,
}
}
// NewSearchQuery creates new search query for the given args.
func NewSearchQuery(start, end int64, tagFilterss [][]TagFilter, maxMetrics int) *SearchQuery {
if maxMetrics <= 0 {
maxMetrics = 2e9
}
return &SearchQuery{
MinTimestamp: start,
MaxTimestamp: end,
TagFilterss: tagFilterss,
MaxMetrics: maxMetrics,
}
}
// TagFilter represents a single tag filter from SearchQuery.
type TagFilter struct {
Key []byte
Value []byte
IsNegative bool
IsRegexp bool
}
// String returns string representation of tf.
func (tf *TagFilter) String() string {
op := tf.getOp()
value := bytesutil.LimitStringLen(string(tf.Value), 60)
if len(tf.Key) == 0 {
return fmt.Sprintf("__name__%s%q", op, value)
}
return fmt.Sprintf("%s%s%q", tf.Key, op, value)
}
func (tf *TagFilter) getOp() string {
if tf.IsNegative {
if tf.IsRegexp {
return "!~"
}
return "!="
}
if tf.IsRegexp {
return "=~"
}
return "="
}
// Marshal appends marshaled tf to dst and returns the result.
func (tf *TagFilter) Marshal(dst []byte) []byte {
dst = encoding.MarshalBytes(dst, tf.Key)
dst = encoding.MarshalBytes(dst, tf.Value)
x := 0
if tf.IsNegative {
x = 2
}
if tf.IsRegexp {
x |= 1
}
dst = append(dst, byte(x))
return dst
}
// Unmarshal unmarshals tf from src and returns the tail.
func (tf *TagFilter) Unmarshal(src []byte) ([]byte, error) {
tail, k, err := encoding.UnmarshalBytes(src)
if err != nil {
return tail, fmt.Errorf("cannot unmarshal Key: %w", err)
}
tf.Key = append(tf.Key[:0], k...)
src = tail
tail, v, err := encoding.UnmarshalBytes(src)
if err != nil {
return tail, fmt.Errorf("cannot unmarshal Value: %w", err)
}
tf.Value = append(tf.Value[:0], v...)
src = tail
if len(src) < 1 {
return src, fmt.Errorf("cannot unmarshal IsNegative+IsRegexp from empty src")
}
x := src[0]
switch x {
case 0:
tf.IsNegative = false
tf.IsRegexp = false
case 1:
tf.IsNegative = false
tf.IsRegexp = true
case 2:
tf.IsNegative = true
tf.IsRegexp = false
case 3:
tf.IsNegative = true
tf.IsRegexp = true
default:
return src, fmt.Errorf("unexpected value for IsNegative+IsRegexp: %d; must be in the range [0..3]", x)
}
src = src[1:]
return src, nil
}
// String returns string representation of the search query.
func (sq *SearchQuery) String() string {
a := make([]string, len(sq.TagFilterss))
for i, tfs := range sq.TagFilterss {
a[i] = tagFiltersToString(tfs)
}
start := TimestampToHumanReadableFormat(sq.MinTimestamp)
end := TimestampToHumanReadableFormat(sq.MaxTimestamp)
return fmt.Sprintf("filters=%s, timeRange=[%s..%s]", a, start, end)
}
func tagFiltersToString(tfs []TagFilter) string {
a := make([]string, len(tfs))
for i, tf := range tfs {
a[i] = tf.String()
}
return "{" + strings.Join(a, ",") + "}"
}
// Marshal appends marshaled sq to dst and returns the result.
func (sq *SearchQuery) Marshal(dst []byte) []byte {
dst = encoding.MarshalVarInt64(dst, sq.MinTimestamp)
dst = encoding.MarshalVarInt64(dst, sq.MaxTimestamp)
dst = encoding.MarshalVarUint64(dst, uint64(len(sq.TagFilterss)))
for _, tagFilters := range sq.TagFilterss {
dst = encoding.MarshalVarUint64(dst, uint64(len(tagFilters)))
for i := range tagFilters {
dst = tagFilters[i].Marshal(dst)
}
}
return dst
}
// Unmarshal unmarshals sq from src and returns the tail.
func (sq *SearchQuery) Unmarshal(src []byte) ([]byte, error) {
tail, minTs, err := encoding.UnmarshalVarInt64(src)
if err != nil {
return src, fmt.Errorf("cannot unmarshal MinTimestamp: %w", err)
}
sq.MinTimestamp = minTs
src = tail
tail, maxTs, err := encoding.UnmarshalVarInt64(src)
if err != nil {
return src, fmt.Errorf("cannot unmarshal MaxTimestamp: %w", err)
}
sq.MaxTimestamp = maxTs
src = tail
tail, tfssCount, err := encoding.UnmarshalVarUint64(src)
if err != nil {
return src, fmt.Errorf("cannot unmarshal the count of TagFilterss: %w", err)
}
if n := int(tfssCount) - cap(sq.TagFilterss); n > 0 {
sq.TagFilterss = append(sq.TagFilterss[:cap(sq.TagFilterss)], make([][]TagFilter, n)...)
}
sq.TagFilterss = sq.TagFilterss[:tfssCount]
src = tail
for i := 0; i < int(tfssCount); i++ {
tail, tfsCount, err := encoding.UnmarshalVarUint64(src)
if err != nil {
return src, fmt.Errorf("cannot unmarshal the count of TagFilters: %w", err)
}
src = tail
tagFilters := sq.TagFilterss[i]
if n := int(tfsCount) - cap(tagFilters); n > 0 {
tagFilters = append(tagFilters[:cap(tagFilters)], make([]TagFilter, n)...)
}
tagFilters = tagFilters[:tfsCount]
for j := 0; j < int(tfsCount); j++ {
tail, err := tagFilters[j].Unmarshal(src)
if err != nil {
return tail, fmt.Errorf("cannot unmarshal TagFilter #%d: %w", j, err)
}
src = tail
}
sq.TagFilterss[i] = tagFilters
}
return src, nil
}
func checkSearchDeadlineAndPace(deadline uint64) error {
if fasttime.UnixTimestamp() > deadline {
return ErrDeadlineExceeded
}
storagepacelimiter.Search.WaitIfNeeded()
return nil
}
const (
paceLimiterFastIterationsMask = 1<<16 - 1
paceLimiterMediumIterationsMask = 1<<14 - 1
paceLimiterSlowIterationsMask = 1<<12 - 1
)