1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796
|
// 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.
package tsdb
import (
"context"
"crypto/rand"
"fmt"
"io"
"math"
"os"
"path/filepath"
"sort"
"time"
"github.com/go-kit/kit/log"
"github.com/go-kit/kit/log/level"
"github.com/oklog/ulid"
"github.com/pkg/errors"
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/prometheus/storage"
"github.com/prometheus/prometheus/tsdb/chunkenc"
"github.com/prometheus/prometheus/tsdb/chunks"
tsdb_errors "github.com/prometheus/prometheus/tsdb/errors"
"github.com/prometheus/prometheus/tsdb/fileutil"
"github.com/prometheus/prometheus/tsdb/index"
"github.com/prometheus/prometheus/tsdb/tombstones"
)
// ExponentialBlockRanges returns the time ranges based on the stepSize.
func ExponentialBlockRanges(minSize int64, steps, stepSize int) []int64 {
ranges := make([]int64, 0, steps)
curRange := minSize
for i := 0; i < steps; i++ {
ranges = append(ranges, curRange)
curRange = curRange * int64(stepSize)
}
return ranges
}
// Compactor provides compaction against an underlying storage
// of time series data.
type Compactor interface {
// Plan returns a set of directories that can be compacted concurrently.
// The directories can be overlapping.
// Results returned when compactions are in progress are undefined.
Plan(dir string) ([]string, error)
// Write persists a Block into a directory.
// No Block is written when resulting Block has 0 samples, and returns empty ulid.ULID{}.
Write(dest string, b BlockReader, mint, maxt int64, parent *BlockMeta) (ulid.ULID, error)
// Compact runs compaction against the provided directories. Must
// only be called concurrently with results of Plan().
// Can optionally pass a list of already open blocks,
// to avoid having to reopen them.
// When resulting Block has 0 samples
// * No block is written.
// * The source dirs are marked Deletable.
// * Returns empty ulid.ULID{}.
Compact(dest string, dirs []string, open []*Block) (ulid.ULID, error)
}
// LeveledCompactor implements the Compactor interface.
type LeveledCompactor struct {
metrics *compactorMetrics
logger log.Logger
ranges []int64
chunkPool chunkenc.Pool
ctx context.Context
}
type compactorMetrics struct {
ran prometheus.Counter
populatingBlocks prometheus.Gauge
overlappingBlocks prometheus.Counter
duration prometheus.Histogram
chunkSize prometheus.Histogram
chunkSamples prometheus.Histogram
chunkRange prometheus.Histogram
}
func newCompactorMetrics(r prometheus.Registerer) *compactorMetrics {
m := &compactorMetrics{}
m.ran = prometheus.NewCounter(prometheus.CounterOpts{
Name: "prometheus_tsdb_compactions_total",
Help: "Total number of compactions that were executed for the partition.",
})
m.populatingBlocks = prometheus.NewGauge(prometheus.GaugeOpts{
Name: "prometheus_tsdb_compaction_populating_block",
Help: "Set to 1 when a block is currently being written to the disk.",
})
m.overlappingBlocks = prometheus.NewCounter(prometheus.CounterOpts{
Name: "prometheus_tsdb_vertical_compactions_total",
Help: "Total number of compactions done on overlapping blocks.",
})
m.duration = prometheus.NewHistogram(prometheus.HistogramOpts{
Name: "prometheus_tsdb_compaction_duration_seconds",
Help: "Duration of compaction runs",
Buckets: prometheus.ExponentialBuckets(1, 2, 10),
})
m.chunkSize = prometheus.NewHistogram(prometheus.HistogramOpts{
Name: "prometheus_tsdb_compaction_chunk_size_bytes",
Help: "Final size of chunks on their first compaction",
Buckets: prometheus.ExponentialBuckets(32, 1.5, 12),
})
m.chunkSamples = prometheus.NewHistogram(prometheus.HistogramOpts{
Name: "prometheus_tsdb_compaction_chunk_samples",
Help: "Final number of samples on their first compaction",
Buckets: prometheus.ExponentialBuckets(4, 1.5, 12),
})
m.chunkRange = prometheus.NewHistogram(prometheus.HistogramOpts{
Name: "prometheus_tsdb_compaction_chunk_range_seconds",
Help: "Final time range of chunks on their first compaction",
Buckets: prometheus.ExponentialBuckets(100, 4, 10),
})
if r != nil {
r.MustRegister(
m.ran,
m.populatingBlocks,
m.overlappingBlocks,
m.duration,
m.chunkRange,
m.chunkSamples,
m.chunkSize,
)
}
return m
}
// NewLeveledCompactor returns a LeveledCompactor.
func NewLeveledCompactor(ctx context.Context, r prometheus.Registerer, l log.Logger, ranges []int64, pool chunkenc.Pool) (*LeveledCompactor, error) {
if len(ranges) == 0 {
return nil, errors.Errorf("at least one range must be provided")
}
if pool == nil {
pool = chunkenc.NewPool()
}
if l == nil {
l = log.NewNopLogger()
}
return &LeveledCompactor{
ranges: ranges,
chunkPool: pool,
logger: l,
metrics: newCompactorMetrics(r),
ctx: ctx,
}, nil
}
type dirMeta struct {
dir string
meta *BlockMeta
}
// Plan returns a list of compactable blocks in the provided directory.
func (c *LeveledCompactor) Plan(dir string) ([]string, error) {
dirs, err := blockDirs(dir)
if err != nil {
return nil, err
}
if len(dirs) < 1 {
return nil, nil
}
var dms []dirMeta
for _, dir := range dirs {
meta, _, err := readMetaFile(dir)
if err != nil {
return nil, err
}
dms = append(dms, dirMeta{dir, meta})
}
return c.plan(dms)
}
func (c *LeveledCompactor) plan(dms []dirMeta) ([]string, error) {
sort.Slice(dms, func(i, j int) bool {
return dms[i].meta.MinTime < dms[j].meta.MinTime
})
res := c.selectOverlappingDirs(dms)
if len(res) > 0 {
return res, nil
}
// No overlapping blocks, do compaction the usual way.
// We do not include a recently created block with max(minTime), so the block which was just created from WAL.
// This gives users a window of a full block size to piece-wise backup new data without having to care about data overlap.
dms = dms[:len(dms)-1]
for _, dm := range c.selectDirs(dms) {
res = append(res, dm.dir)
}
if len(res) > 0 {
return res, nil
}
// Compact any blocks with big enough time range that have >5% tombstones.
for i := len(dms) - 1; i >= 0; i-- {
meta := dms[i].meta
if meta.MaxTime-meta.MinTime < c.ranges[len(c.ranges)/2] {
// If the block is entirely deleted, then we don't care about the block being big enough.
// TODO: This is assuming single tombstone is for distinct series, which might be no true.
if meta.Stats.NumTombstones > 0 && meta.Stats.NumTombstones >= meta.Stats.NumSeries {
return []string{dms[i].dir}, nil
}
break
}
if float64(meta.Stats.NumTombstones)/float64(meta.Stats.NumSeries+1) > 0.05 {
return []string{dms[i].dir}, nil
}
}
return nil, nil
}
// selectDirs returns the dir metas that should be compacted into a single new block.
// If only a single block range is configured, the result is always nil.
func (c *LeveledCompactor) selectDirs(ds []dirMeta) []dirMeta {
if len(c.ranges) < 2 || len(ds) < 1 {
return nil
}
highTime := ds[len(ds)-1].meta.MinTime
for _, iv := range c.ranges[1:] {
parts := splitByRange(ds, iv)
if len(parts) == 0 {
continue
}
Outer:
for _, p := range parts {
// Do not select the range if it has a block whose compaction failed.
for _, dm := range p {
if dm.meta.Compaction.Failed {
continue Outer
}
}
mint := p[0].meta.MinTime
maxt := p[len(p)-1].meta.MaxTime
// Pick the range of blocks if it spans the full range (potentially with gaps)
// or is before the most recent block.
// This ensures we don't compact blocks prematurely when another one of the same
// size still fits in the range.
if (maxt-mint == iv || maxt <= highTime) && len(p) > 1 {
return p
}
}
}
return nil
}
// selectOverlappingDirs returns all dirs with overlapping time ranges.
// It expects sorted input by mint and returns the overlapping dirs in the same order as received.
func (c *LeveledCompactor) selectOverlappingDirs(ds []dirMeta) []string {
if len(ds) < 2 {
return nil
}
var overlappingDirs []string
globalMaxt := ds[0].meta.MaxTime
for i, d := range ds[1:] {
if d.meta.MinTime < globalMaxt {
if len(overlappingDirs) == 0 { // When it is the first overlap, need to add the last one as well.
overlappingDirs = append(overlappingDirs, ds[i].dir)
}
overlappingDirs = append(overlappingDirs, d.dir)
} else if len(overlappingDirs) > 0 {
break
}
if d.meta.MaxTime > globalMaxt {
globalMaxt = d.meta.MaxTime
}
}
return overlappingDirs
}
// splitByRange splits the directories by the time range. The range sequence starts at 0.
//
// For example, if we have blocks [0-10, 10-20, 50-60, 90-100] and the split range tr is 30
// it returns [0-10, 10-20], [50-60], [90-100].
func splitByRange(ds []dirMeta, tr int64) [][]dirMeta {
var splitDirs [][]dirMeta
for i := 0; i < len(ds); {
var (
group []dirMeta
t0 int64
m = ds[i].meta
)
// Compute start of aligned time range of size tr closest to the current block's start.
if m.MinTime >= 0 {
t0 = tr * (m.MinTime / tr)
} else {
t0 = tr * ((m.MinTime - tr + 1) / tr)
}
// Skip blocks that don't fall into the range. This can happen via mis-alignment or
// by being the multiple of the intended range.
if m.MaxTime > t0+tr {
i++
continue
}
// Add all dirs to the current group that are within [t0, t0+tr].
for ; i < len(ds); i++ {
// Either the block falls into the next range or doesn't fit at all (checked above).
if ds[i].meta.MaxTime > t0+tr {
break
}
group = append(group, ds[i])
}
if len(group) > 0 {
splitDirs = append(splitDirs, group)
}
}
return splitDirs
}
// CompactBlockMetas merges many block metas into one, combining it's source blocks together
// and adjusting compaction level.
func CompactBlockMetas(uid ulid.ULID, blocks ...*BlockMeta) *BlockMeta {
res := &BlockMeta{
ULID: uid,
MinTime: blocks[0].MinTime,
}
sources := map[ulid.ULID]struct{}{}
// For overlapping blocks, the Maxt can be
// in any block so we track it globally.
maxt := int64(math.MinInt64)
for _, b := range blocks {
if b.MaxTime > maxt {
maxt = b.MaxTime
}
if b.Compaction.Level > res.Compaction.Level {
res.Compaction.Level = b.Compaction.Level
}
for _, s := range b.Compaction.Sources {
sources[s] = struct{}{}
}
res.Compaction.Parents = append(res.Compaction.Parents, BlockDesc{
ULID: b.ULID,
MinTime: b.MinTime,
MaxTime: b.MaxTime,
})
}
res.Compaction.Level++
for s := range sources {
res.Compaction.Sources = append(res.Compaction.Sources, s)
}
sort.Slice(res.Compaction.Sources, func(i, j int) bool {
return res.Compaction.Sources[i].Compare(res.Compaction.Sources[j]) < 0
})
res.MaxTime = maxt
return res
}
// Compact creates a new block in the compactor's directory from the blocks in the
// provided directories.
func (c *LeveledCompactor) Compact(dest string, dirs []string, open []*Block) (uid ulid.ULID, err error) {
var (
blocks []BlockReader
bs []*Block
metas []*BlockMeta
uids []string
)
start := time.Now()
for _, d := range dirs {
meta, _, err := readMetaFile(d)
if err != nil {
return uid, err
}
var b *Block
// Use already open blocks if we can, to avoid
// having the index data in memory twice.
for _, o := range open {
if meta.ULID == o.Meta().ULID {
b = o
break
}
}
if b == nil {
var err error
b, err = OpenBlock(c.logger, d, c.chunkPool)
if err != nil {
return uid, err
}
defer b.Close()
}
metas = append(metas, meta)
blocks = append(blocks, b)
bs = append(bs, b)
uids = append(uids, meta.ULID.String())
}
uid = ulid.MustNew(ulid.Now(), rand.Reader)
meta := CompactBlockMetas(uid, metas...)
err = c.write(dest, meta, blocks...)
if err == nil {
if meta.Stats.NumSamples == 0 {
for _, b := range bs {
b.meta.Compaction.Deletable = true
n, err := writeMetaFile(c.logger, b.dir, &b.meta)
if err != nil {
level.Error(c.logger).Log(
"msg", "Failed to write 'Deletable' to meta file after compaction",
"ulid", b.meta.ULID,
)
}
b.numBytesMeta = n
}
uid = ulid.ULID{}
level.Info(c.logger).Log(
"msg", "compact blocks resulted in empty block",
"count", len(blocks),
"sources", fmt.Sprintf("%v", uids),
"duration", time.Since(start),
)
} else {
level.Info(c.logger).Log(
"msg", "compact blocks",
"count", len(blocks),
"mint", meta.MinTime,
"maxt", meta.MaxTime,
"ulid", meta.ULID,
"sources", fmt.Sprintf("%v", uids),
"duration", time.Since(start),
)
}
return uid, nil
}
errs := tsdb_errors.NewMulti(err)
if err != context.Canceled {
for _, b := range bs {
if err := b.setCompactionFailed(); err != nil {
errs.Add(errors.Wrapf(err, "setting compaction failed for block: %s", b.Dir()))
}
}
}
return uid, errs.Err()
}
func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64, parent *BlockMeta) (ulid.ULID, error) {
start := time.Now()
uid := ulid.MustNew(ulid.Now(), rand.Reader)
meta := &BlockMeta{
ULID: uid,
MinTime: mint,
MaxTime: maxt,
}
meta.Compaction.Level = 1
meta.Compaction.Sources = []ulid.ULID{uid}
if parent != nil {
meta.Compaction.Parents = []BlockDesc{
{ULID: parent.ULID, MinTime: parent.MinTime, MaxTime: parent.MaxTime},
}
}
err := c.write(dest, meta, b)
if err != nil {
return uid, err
}
if meta.Stats.NumSamples == 0 {
level.Info(c.logger).Log(
"msg", "write block resulted in empty block",
"mint", meta.MinTime,
"maxt", meta.MaxTime,
"duration", time.Since(start),
)
return ulid.ULID{}, nil
}
level.Info(c.logger).Log(
"msg", "write block",
"mint", meta.MinTime,
"maxt", meta.MaxTime,
"ulid", meta.ULID,
"duration", time.Since(start),
)
return uid, nil
}
// instrumentedChunkWriter is used for level 1 compactions to record statistics
// about compacted chunks.
type instrumentedChunkWriter struct {
ChunkWriter
size prometheus.Histogram
samples prometheus.Histogram
trange prometheus.Histogram
}
func (w *instrumentedChunkWriter) WriteChunks(chunks ...chunks.Meta) error {
for _, c := range chunks {
w.size.Observe(float64(len(c.Chunk.Bytes())))
w.samples.Observe(float64(c.Chunk.NumSamples()))
w.trange.Observe(float64(c.MaxTime - c.MinTime))
}
return w.ChunkWriter.WriteChunks(chunks...)
}
// write creates a new block that is the union of the provided blocks into dir.
func (c *LeveledCompactor) write(dest string, meta *BlockMeta, blocks ...BlockReader) (err error) {
dir := filepath.Join(dest, meta.ULID.String())
tmp := dir + tmpForCreationBlockDirSuffix
var closers []io.Closer
defer func(t time.Time) {
err = tsdb_errors.NewMulti(err, tsdb_errors.CloseAll(closers)).Err()
// RemoveAll returns no error when tmp doesn't exist so it is safe to always run it.
if err := os.RemoveAll(tmp); err != nil {
level.Error(c.logger).Log("msg", "removed tmp folder after failed compaction", "err", err.Error())
}
c.metrics.ran.Inc()
c.metrics.duration.Observe(time.Since(t).Seconds())
}(time.Now())
if err = os.RemoveAll(tmp); err != nil {
return err
}
if err = os.MkdirAll(tmp, 0777); err != nil {
return err
}
// Populate chunk and index files into temporary directory with
// data of all blocks.
var chunkw ChunkWriter
chunkw, err = chunks.NewWriter(chunkDir(tmp))
if err != nil {
return errors.Wrap(err, "open chunk writer")
}
closers = append(closers, chunkw)
// Record written chunk sizes on level 1 compactions.
if meta.Compaction.Level == 1 {
chunkw = &instrumentedChunkWriter{
ChunkWriter: chunkw,
size: c.metrics.chunkSize,
samples: c.metrics.chunkSamples,
trange: c.metrics.chunkRange,
}
}
indexw, err := index.NewWriter(c.ctx, filepath.Join(tmp, indexFilename))
if err != nil {
return errors.Wrap(err, "open index writer")
}
closers = append(closers, indexw)
if err := c.populateBlock(blocks, meta, indexw, chunkw); err != nil {
return errors.Wrap(err, "populate block")
}
select {
case <-c.ctx.Done():
return c.ctx.Err()
default:
}
// We are explicitly closing them here to check for error even
// though these are covered under defer. This is because in Windows,
// you cannot delete these unless they are closed and the defer is to
// make sure they are closed if the function exits due to an error above.
errs := tsdb_errors.NewMulti()
for _, w := range closers {
errs.Add(w.Close())
}
closers = closers[:0] // Avoid closing the writers twice in the defer.
if errs.Err() != nil {
return errs.Err()
}
// Populated block is empty, so exit early.
if meta.Stats.NumSamples == 0 {
return nil
}
if _, err = writeMetaFile(c.logger, tmp, meta); err != nil {
return errors.Wrap(err, "write merged meta")
}
// Create an empty tombstones file.
if _, err := tombstones.WriteFile(c.logger, tmp, tombstones.NewMemTombstones()); err != nil {
return errors.Wrap(err, "write new tombstones file")
}
df, err := fileutil.OpenDir(tmp)
if err != nil {
return errors.Wrap(err, "open temporary block dir")
}
defer func() {
if df != nil {
df.Close()
}
}()
if err := df.Sync(); err != nil {
return errors.Wrap(err, "sync temporary dir file")
}
// Close temp dir before rename block dir (for windows platform).
if err = df.Close(); err != nil {
return errors.Wrap(err, "close temporary dir")
}
df = nil
// Block successfully written, make it visible in destination dir by moving it from tmp one.
if err := fileutil.Replace(tmp, dir); err != nil {
return errors.Wrap(err, "rename block dir")
}
return nil
}
// populateBlock fills the index and chunk writers with new data gathered as the union
// of the provided blocks. It returns meta information for the new block.
// It expects sorted blocks input by mint.
func (c *LeveledCompactor) populateBlock(blocks []BlockReader, meta *BlockMeta, indexw IndexWriter, chunkw ChunkWriter) (err error) {
if len(blocks) == 0 {
return errors.New("cannot populate block from no readers")
}
var (
sets []storage.ChunkSeriesSet
symbols index.StringIter
closers []io.Closer
overlapping bool
)
defer func() {
errs := tsdb_errors.NewMulti(err)
if cerr := tsdb_errors.CloseAll(closers); cerr != nil {
errs.Add(errors.Wrap(cerr, "close"))
}
err = errs.Err()
c.metrics.populatingBlocks.Set(0)
}()
c.metrics.populatingBlocks.Set(1)
globalMaxt := blocks[0].Meta().MaxTime
for i, b := range blocks {
select {
case <-c.ctx.Done():
return c.ctx.Err()
default:
}
if !overlapping {
if i > 0 && b.Meta().MinTime < globalMaxt {
c.metrics.overlappingBlocks.Inc()
overlapping = true
level.Info(c.logger).Log("msg", "Found overlapping blocks during compaction", "ulid", meta.ULID)
}
if b.Meta().MaxTime > globalMaxt {
globalMaxt = b.Meta().MaxTime
}
}
indexr, err := b.Index()
if err != nil {
return errors.Wrapf(err, "open index reader for block %+v", b.Meta())
}
closers = append(closers, indexr)
chunkr, err := b.Chunks()
if err != nil {
return errors.Wrapf(err, "open chunk reader for block %+v", b.Meta())
}
closers = append(closers, chunkr)
tombsr, err := b.Tombstones()
if err != nil {
return errors.Wrapf(err, "open tombstone reader for block %+v", b.Meta())
}
closers = append(closers, tombsr)
k, v := index.AllPostingsKey()
all, err := indexr.Postings(k, v)
if err != nil {
return err
}
all = indexr.SortedPostings(all)
// Blocks meta is half open: [min, max), so subtract 1 to ensure we don't hold samples with exact meta.MaxTime timestamp.
sets = append(sets, newBlockChunkSeriesSet(indexr, chunkr, tombsr, all, meta.MinTime, meta.MaxTime-1))
syms := indexr.Symbols()
if i == 0 {
symbols = syms
continue
}
symbols = NewMergedStringIter(symbols, syms)
}
for symbols.Next() {
if err := indexw.AddSymbol(symbols.At()); err != nil {
return errors.Wrap(err, "add symbol")
}
}
if symbols.Err() != nil {
return errors.Wrap(symbols.Err(), "next symbol")
}
var (
ref = uint64(0)
chks []chunks.Meta
)
set := sets[0]
if len(sets) > 1 {
// Merge series using compacting chunk series merger.
set = storage.NewMergeChunkSeriesSet(sets, storage.NewCompactingChunkSeriesMerger(storage.ChainedSeriesMerge))
}
// Iterate over all sorted chunk series.
for set.Next() {
select {
case <-c.ctx.Done():
return c.ctx.Err()
default:
}
s := set.At()
chksIter := s.Iterator()
chks = chks[:0]
for chksIter.Next() {
// We are not iterating in streaming way over chunk as it's more efficient to do bulk write for index and
// chunk file purposes.
chks = append(chks, chksIter.At())
}
if chksIter.Err() != nil {
return errors.Wrap(chksIter.Err(), "chunk iter")
}
// Skip the series with all deleted chunks.
if len(chks) == 0 {
continue
}
if err := chunkw.WriteChunks(chks...); err != nil {
return errors.Wrap(err, "write chunks")
}
if err := indexw.AddSeries(ref, s.Labels(), chks...); err != nil {
return errors.Wrap(err, "add series")
}
meta.Stats.NumChunks += uint64(len(chks))
meta.Stats.NumSeries++
for _, chk := range chks {
meta.Stats.NumSamples += uint64(chk.Chunk.NumSamples())
}
for _, chk := range chks {
if err := c.chunkPool.Put(chk.Chunk); err != nil {
return errors.Wrap(err, "put chunk")
}
}
ref++
}
if set.Err() != nil {
return errors.Wrap(set.Err(), "iterate compaction set")
}
return nil
}
|