You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
prometheus/tsdb/chunks/head_chunks_test.go

584 lines
16 KiB

// Copyright 2020 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 chunks
import (
"encoding/binary"
"errors"
"math/rand"
"os"
"strconv"
"sync"
"testing"
"time"
"github.com/stretchr/testify/require"
"github.com/prometheus/prometheus/tsdb/chunkenc"
)
var writeQueueSize int
func TestMain(m *testing.M) {
// Run all tests with the chunk write queue disabled.
writeQueueSize = 0
exitVal := m.Run()
if exitVal != 0 {
os.Exit(exitVal)
}
// Re-run all tests with the chunk write queue size of 1e6.
writeQueueSize = 1000000
exitVal = m.Run()
os.Exit(exitVal)
}
func TestChunkDiskMapper_WriteChunk_Chunk_IterateChunks(t *testing.T) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw := createChunkDiskMapper(t, "")
defer func() {
require.NoError(t, hrw.Close())
}()
expectedBytes := []byte{}
nextChunkOffset := uint64(HeadChunkFileHeaderSize)
chkCRC32 := newCRC32()
type expectedDataType struct {
seriesRef HeadSeriesRef
chunkRef ChunkDiskMapperRef
mint, maxt int64
numSamples uint16
chunk chunkenc.Chunk
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
isOOO bool
}
expectedData := []expectedDataType{}
var buf [MaxHeadChunkMetaSize]byte
totalChunks := 0
var firstFileName string
for hrw.curFileSequence < 3 || hrw.chkWriter.Buffered() == 0 {
addChunks := func(numChunks int) {
for i := 0; i < numChunks; i++ {
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
seriesRef, chkRef, mint, maxt, chunk, isOOO := createChunk(t, totalChunks, hrw)
totalChunks++
expectedData = append(expectedData, expectedDataType{
seriesRef: seriesRef,
mint: mint,
maxt: maxt,
chunkRef: chkRef,
chunk: chunk,
numSamples: uint16(chunk.NumSamples()),
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
isOOO: isOOO,
})
if hrw.curFileSequence != 1 {
// We are checking for bytes written only for the first file.
continue
}
// Calculating expected bytes written on disk for first file.
firstFileName = hrw.curFile.Name()
require.Equal(t, newChunkDiskMapperRef(1, nextChunkOffset), chkRef)
bytesWritten := 0
chkCRC32.Reset()
binary.BigEndian.PutUint64(buf[bytesWritten:], uint64(seriesRef))
bytesWritten += SeriesRefSize
binary.BigEndian.PutUint64(buf[bytesWritten:], uint64(mint))
bytesWritten += MintMaxtSize
binary.BigEndian.PutUint64(buf[bytesWritten:], uint64(maxt))
bytesWritten += MintMaxtSize
enc := chunk.Encoding()
if isOOO {
enc = hrw.ApplyOutOfOrderMask(enc)
}
buf[bytesWritten] = byte(enc)
bytesWritten += ChunkEncodingSize
n := binary.PutUvarint(buf[bytesWritten:], uint64(len(chunk.Bytes())))
bytesWritten += n
expectedBytes = append(expectedBytes, buf[:bytesWritten]...)
_, err := chkCRC32.Write(buf[:bytesWritten])
require.NoError(t, err)
expectedBytes = append(expectedBytes, chunk.Bytes()...)
_, err = chkCRC32.Write(chunk.Bytes())
require.NoError(t, err)
expectedBytes = append(expectedBytes, chkCRC32.Sum(nil)...)
// += seriesRef, mint, maxt, encoding, chunk data len, chunk data, CRC.
nextChunkOffset += SeriesRefSize + 2*MintMaxtSize + ChunkEncodingSize + uint64(n) + uint64(len(chunk.Bytes())) + CRCSize
}
}
addChunks(100)
hrw.CutNewFile()
addChunks(10) // For chunks in in-memory buffer.
}
// Checking on-disk bytes for the first file.
require.Len(t, hrw.mmappedChunkFiles, 3, "expected 3 mmapped files, got %d", len(hrw.mmappedChunkFiles))
require.Equal(t, len(hrw.mmappedChunkFiles), len(hrw.closers))
actualBytes, err := os.ReadFile(firstFileName)
require.NoError(t, err)
// Check header of the segment file.
require.Equal(t, MagicHeadChunks, int(binary.BigEndian.Uint32(actualBytes[0:MagicChunksSize])))
require.Equal(t, chunksFormatV1, int(actualBytes[MagicChunksSize]))
// Remaining chunk data.
fileEnd := HeadChunkFileHeaderSize + len(expectedBytes)
require.Equal(t, expectedBytes, actualBytes[HeadChunkFileHeaderSize:fileEnd])
// Testing reading of chunks.
for _, exp := range expectedData {
actChunk, err := hrw.Chunk(exp.chunkRef)
require.NoError(t, err)
require.Equal(t, exp.chunk.Bytes(), actChunk.Bytes())
}
// Testing IterateAllChunks method.
dir := hrw.dir.Name()
require.NoError(t, hrw.Close())
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw = createChunkDiskMapper(t, dir)
idx := 0
require.NoError(t, hrw.IterateAllChunks(func(seriesRef HeadSeriesRef, chunkRef ChunkDiskMapperRef, mint, maxt int64, numSamples uint16, encoding chunkenc.Encoding, isOOO bool) error {
t.Helper()
expData := expectedData[idx]
require.Equal(t, expData.seriesRef, seriesRef)
require.Equal(t, expData.chunkRef, chunkRef)
require.Equal(t, expData.maxt, maxt)
require.Equal(t, expData.maxt, maxt)
require.Equal(t, expData.numSamples, numSamples)
require.Equal(t, expData.isOOO, isOOO)
actChunk, err := hrw.Chunk(expData.chunkRef)
require.NoError(t, err)
require.Equal(t, expData.chunk.Bytes(), actChunk.Bytes())
idx++
return nil
}))
require.Len(t, expectedData, idx)
}
// TestChunkDiskMapper_Truncate tests
// * If truncation is happening properly based on the time passed.
// * The active file is not deleted even if the passed time makes it eligible to be deleted.
// * Non-empty current file leads to creation of another file after truncation.
func TestChunkDiskMapper_Truncate(t *testing.T) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw := createChunkDiskMapper(t, "")
defer func() {
require.NoError(t, hrw.Close())
}()
timeRange := 0
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
addChunk := func() {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
t.Helper()
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
step := 100
mint, maxt := timeRange+1, timeRange+step-1
var err error
awaitCb := make(chan struct{})
hrw.WriteChunk(1, int64(mint), int64(maxt), randomChunk(t), false, func(cbErr error) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
err = cbErr
close(awaitCb)
})
<-awaitCb
require.NoError(t, err)
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
timeRange += step
}
verifyFiles := func(remainingFiles []int) {
t.Helper()
files, err := os.ReadDir(hrw.dir.Name())
require.NoError(t, err)
require.Equal(t, len(remainingFiles), len(files), "files on disk")
require.Equal(t, len(remainingFiles), len(hrw.mmappedChunkFiles), "hrw.mmappedChunkFiles")
require.Equal(t, len(remainingFiles), len(hrw.closers), "closers")
for _, i := range remainingFiles {
_, ok := hrw.mmappedChunkFiles[i]
require.True(t, ok)
}
}
// Create segments 1 to 7.
for i := 1; i <= 7; i++ {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw.CutNewFile()
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
addChunk()
}
verifyFiles([]int{1, 2, 3, 4, 5, 6, 7})
// Truncating files.
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
require.NoError(t, hrw.Truncate(3))
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
// Add a chunk to trigger cutting of new file.
addChunk()
verifyFiles([]int{3, 4, 5, 6, 7, 8})
dir := hrw.dir.Name()
require.NoError(t, hrw.Close())
// Restarted.
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw = createChunkDiskMapper(t, dir)
verifyFiles([]int{3, 4, 5, 6, 7, 8})
// New file is created after restart even if last file was empty.
addChunk()
verifyFiles([]int{3, 4, 5, 6, 7, 8, 9})
// Truncating files after restart.
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
require.NoError(t, hrw.Truncate(6))
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
verifyFiles([]int{6, 7, 8, 9})
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
// Truncating a second time without adding a chunk shouldn't create a new file.
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
require.NoError(t, hrw.Truncate(6))
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
verifyFiles([]int{6, 7, 8, 9})
// Add a chunk to trigger cutting of new file.
addChunk()
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
verifyFiles([]int{6, 7, 8, 9, 10})
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
// Truncation by file number.
require.NoError(t, hrw.Truncate(8))
verifyFiles([]int{8, 9, 10})
// Truncating till current time should not delete the current active file.
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
require.NoError(t, hrw.Truncate(10))
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
// Add a chunk to trigger cutting of new file.
addChunk()
verifyFiles([]int{10, 11}) // One file is the previously active file and one currently created.
}
// TestChunkDiskMapper_Truncate_PreservesFileSequence tests that truncation doesn't poke
// holes into the file sequence, even if there are empty files in between non-empty files.
// This test exposes https://github.com/prometheus/prometheus/issues/7412 where the truncation
// simply deleted all empty files instead of stopping once it encountered a non-empty file.
func TestChunkDiskMapper_Truncate_PreservesFileSequence(t *testing.T) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw := createChunkDiskMapper(t, "")
defer func() {
require.NoError(t, hrw.Close())
}()
timeRange := 0
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
addChunk := func() {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
t.Helper()
awaitCb := make(chan struct{})
step := 100
mint, maxt := timeRange+1, timeRange+step-1
hrw.WriteChunk(1, int64(mint), int64(maxt), randomChunk(t), false, func(err error) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
close(awaitCb)
require.NoError(t, err)
})
<-awaitCb
timeRange += step
}
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
emptyFile := func() {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
t.Helper()
_, _, err := hrw.cut()
require.NoError(t, err)
hrw.evtlPosMtx.Lock()
hrw.evtlPos.toNewFile()
hrw.evtlPosMtx.Unlock()
}
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
nonEmptyFile := func() {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
t.Helper()
emptyFile()
addChunk()
}
addChunk() // 1. Created with the first chunk.
nonEmptyFile() // 2.
nonEmptyFile() // 3.
emptyFile() // 4.
nonEmptyFile() // 5.
emptyFile() // 6.
verifyFiles := func(remainingFiles []int) {
t.Helper()
files, err := os.ReadDir(hrw.dir.Name())
require.NoError(t, err)
require.Equal(t, len(remainingFiles), len(files), "files on disk")
require.Equal(t, len(remainingFiles), len(hrw.mmappedChunkFiles), "hrw.mmappedChunkFiles")
require.Equal(t, len(remainingFiles), len(hrw.closers), "closers")
for _, i := range remainingFiles {
_, ok := hrw.mmappedChunkFiles[i]
require.True(t, ok, "remaining file %d not in hrw.mmappedChunkFiles", i)
}
}
verifyFiles([]int{1, 2, 3, 4, 5, 6})
// Truncating files till 2. It should not delete anything after 3 (inclusive)
// though files 4 and 6 are empty.
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
require.NoError(t, hrw.Truncate(3))
verifyFiles([]int{3, 4, 5, 6})
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
// Add chunk, so file 6 is not empty anymore.
addChunk()
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
verifyFiles([]int{3, 4, 5, 6})
// Truncating till file 3 should also delete file 4, because it is empty.
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
require.NoError(t, hrw.Truncate(5))
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
addChunk()
verifyFiles([]int{5, 6, 7})
dir := hrw.dir.Name()
require.NoError(t, hrw.Close())
// Restarting checks for unsequential files.
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw = createChunkDiskMapper(t, dir)
verifyFiles([]int{5, 6, 7})
}
func TestChunkDiskMapper_Truncate_WriteQueueRaceCondition(t *testing.T) {
hrw := createChunkDiskMapper(t, "")
t.Cleanup(func() {
require.NoError(t, hrw.Close())
})
// This test should only run when the queue is enabled.
if hrw.writeQueue == nil {
t.Skip("This test should only run when the queue is enabled")
}
// Add an artificial delay in the writeChunk function to easily trigger the race condition.
origWriteChunk := hrw.writeQueue.writeChunk
hrw.writeQueue.writeChunk = func(seriesRef HeadSeriesRef, mint, maxt int64, chk chunkenc.Chunk, ref ChunkDiskMapperRef, isOOO, cutFile bool) error {
time.Sleep(100 * time.Millisecond)
return origWriteChunk(seriesRef, mint, maxt, chk, ref, isOOO, cutFile)
}
wg := sync.WaitGroup{}
wg.Add(2)
// Write a chunk. Since the queue is enabled, the chunk will be written asynchronously (with the artificial delay).
ref := hrw.WriteChunk(1, 0, 10, randomChunk(t), false, func(err error) {
defer wg.Done()
require.NoError(t, err)
})
seq, _ := ref.Unpack()
require.Equal(t, 1, seq)
// Truncate, simulating that all chunks from segment files before 1 can be dropped.
require.NoError(t, hrw.Truncate(1))
// Request to cut a new file when writing the next chunk. If there's a race condition, cutting a new file will
// allow us to detect there's actually an issue with the sequence number (because it's checked when a new segment
// file is created).
hrw.CutNewFile()
// Write another chunk. This will cut a new file.
ref = hrw.WriteChunk(1, 0, 10, randomChunk(t), false, func(err error) {
defer wg.Done()
require.NoError(t, err)
})
seq, _ = ref.Unpack()
require.Equal(t, 2, seq)
wg.Wait()
}
// TestHeadReadWriter_TruncateAfterIterateChunksError tests for
// https://github.com/prometheus/prometheus/issues/7753
func TestHeadReadWriter_TruncateAfterFailedIterateChunks(t *testing.T) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw := createChunkDiskMapper(t, "")
defer func() {
require.NoError(t, hrw.Close())
}()
// Write a chunks to iterate on it later.
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
var err error
awaitCb := make(chan struct{})
hrw.WriteChunk(1, 0, 1000, randomChunk(t), false, func(cbErr error) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
err = cbErr
close(awaitCb)
})
<-awaitCb
require.NoError(t, err)
dir := hrw.dir.Name()
require.NoError(t, hrw.Close())
// Restarting to recreate https://github.com/prometheus/prometheus/issues/7753.
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw = createChunkDiskMapper(t, dir)
// Forcefully failing IterateAllChunks.
require.Error(t, hrw.IterateAllChunks(func(_ HeadSeriesRef, _ ChunkDiskMapperRef, _, _ int64, _ uint16, _ chunkenc.Encoding, _ bool) error {
return errors.New("random error")
}))
// Truncation call should not return error after IterateAllChunks fails.
require.NoError(t, hrw.Truncate(2000))
}
func TestHeadReadWriter_ReadRepairOnEmptyLastFile(t *testing.T) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
hrw := createChunkDiskMapper(t, "")
timeRange := 0
addChunk := func() {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
t.Helper()
step := 100
mint, maxt := timeRange+1, timeRange+step-1
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
var err error
awaitCb := make(chan struct{})
hrw.WriteChunk(1, int64(mint), int64(maxt), randomChunk(t), false, func(cbErr error) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
err = cbErr
close(awaitCb)
})
<-awaitCb
require.NoError(t, err)
timeRange += step
}
nonEmptyFile := func() {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
t.Helper()
hrw.CutNewFile()
addChunk()
}
addChunk() // 1. Created with the first chunk.
nonEmptyFile() // 2.
nonEmptyFile() // 3.
require.Len(t, hrw.mmappedChunkFiles, 3)
lastFile := 0
for idx := range hrw.mmappedChunkFiles {
if idx > lastFile {
lastFile = idx
}
}
require.Equal(t, 3, lastFile)
dir := hrw.dir.Name()
require.NoError(t, hrw.Close())
writeCorruptLastFile := func(b []byte) {
fname := segmentFile(dir, lastFile+1)
f, err := os.OpenFile(fname, os.O_WRONLY|os.O_CREATE, 0o666)
require.NoError(t, err)
_, err = f.Write(b)
require.NoError(t, err)
require.NoError(t, f.Sync())
stat, err := f.Stat()
require.NoError(t, err)
require.Equal(t, int64(len(b)), stat.Size())
require.NoError(t, f.Close())
}
checkRepair := func() {
// Open chunk disk mapper again, corrupt file should be removed.
hrw = createChunkDiskMapper(t, dir)
// Removed from memory.
require.Len(t, hrw.mmappedChunkFiles, 3)
for idx := range hrw.mmappedChunkFiles {
require.LessOrEqual(t, idx, lastFile, "file index is bigger than previous last file")
}
// Removed even from disk.
files, err := os.ReadDir(dir)
require.NoError(t, err)
require.Len(t, files, 3)
for _, fi := range files {
seq, err := strconv.ParseUint(fi.Name(), 10, 64)
require.NoError(t, err)
require.LessOrEqual(t, seq, uint64(lastFile), "file index on disk is bigger than previous last file")
}
require.NoError(t, hrw.Close())
}
// Write an empty last file mimicking an abrupt shutdown on file creation.
writeCorruptLastFile(nil)
// Removes empty last file.
checkRepair()
// Write another empty last file with 0 bytes.
writeCorruptLastFile([]byte{0, 0, 0, 0, 0, 0, 0, 0})
// Removes the 0 filled last file.
checkRepair()
// Write another corrupt file with less than 4 bytes (size of magic number).
writeCorruptLastFile([]byte{1, 2})
// Removes the partial file.
checkRepair()
// Check that it does not delete a valid last file.
checkRepair()
}
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
func createChunkDiskMapper(t *testing.T, dir string) *ChunkDiskMapper {
if dir == "" {
dir = t.TempDir()
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
}
hrw, err := NewChunkDiskMapper(nil, dir, chunkenc.NewPool(), DefaultWriteBufferSize, writeQueueSize)
require.NoError(t, err)
require.False(t, hrw.fileMaxtSet)
require.NoError(t, hrw.IterateAllChunks(func(_ HeadSeriesRef, _ ChunkDiskMapperRef, _, _ int64, _ uint16, _ chunkenc.Encoding, _ bool) error {
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
return nil
}))
require.True(t, hrw.fileMaxtSet)
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
return hrw
}
func randomChunk(t *testing.T) chunkenc.Chunk {
chunk := chunkenc.NewXORChunk()
length := rand.Int() % 120
app, err := chunk.Appender()
require.NoError(t, err)
for i := 0; i < length; i++ {
app.Append(rand.Int63(), rand.Float64())
}
return chunk
}
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
func createChunk(t *testing.T, idx int, hrw *ChunkDiskMapper) (seriesRef HeadSeriesRef, chunkRef ChunkDiskMapperRef, mint, maxt int64, chunk chunkenc.Chunk, isOOO bool) {
var err error
seriesRef = HeadSeriesRef(rand.Int63())
mint = int64((idx)*1000 + 1)
maxt = int64((idx + 1) * 1000)
chunk = randomChunk(t)
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
awaitCb := make(chan struct{})
Add out-of-order sample support to the TSDB (#11075) * Introduce out-of-order TSDB support This implementation is based on this design doc: https://docs.google.com/document/d/1Kppm7qL9C-BJB1j6yb6-9ObG3AbdZnFUBYPNNWwDBYM/edit?usp=sharing This commit adds support to accept out-of-order ("OOO") sample into the TSDB up to a configurable time allowance. If OOO is enabled, overlapping querying are automatically enabled. Most of the additions have been borrowed from https://github.com/grafana/mimir-prometheus/ Here is the list ist of the original commits cherry picked from mimir-prometheus into this branch: - 4b2198d7ec47d50989b7c2df66b7b207c32f7f6e - 2836e5513f1bc591535a859f5d41154a75e7c6bc - 00b379c3a5b1ec3799699b6242f300a2b3ea30f0 - ff0dc757587cada63ca948d2d5eb00bf090d63e0 - a632c73352a7e39d60b445700beb47d691549c3e - c6f3d4ab339ab80bbbce74c9946237ced01f0509 - 5e8406a1d4a50d0052bbee83e28ca3b3371408aa - abde1e0ba128936b9eb0224ee1551e56216ebd4a - e70e7698897bb03860bee0467c733fa44e14c9bd - df59320886e03a555d379ac4b0b3130f661407e0 Co-authored-by: Jesus Vazquez <jesus.vazquez@grafana.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * gofumpt files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add license header to missing files Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO tests due to existing chunk disk mapper implementation Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix truncate int overflow Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Add Sync method to the WAL and update tests Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * remove useless sync Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Update minOOOTime after truncating Head * Update minOOOTime after truncating Head Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix lint Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Add a unit test Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Load OutOfOrderTimeWindow only once per appender Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix OOO Head LabelValues and PostingsForMatchers Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix replay of OOO mmap chunks Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Remove unnecessary err check Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Prevent panic with ApplyConfig Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run OOO compaction after restart if there is OOO data from WBL Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Apply Bartek's suggestions Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Refactor OOO compaction Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address comments and TODOs - Added a comment explaining why we need the allow overlapping compaction toggle - Clarified TSDBConfig OutOfOrderTimeWindow doc - Added an owner to all the TODOs in the code Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Run go format Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix remaining review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix tests Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Change wbl reference when truncating ooo in TestHeadMinOOOTimeUpdate Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> * Fix TestWBLAndMmapReplay test failure on windows Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Address most of the feedback Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Refactor the block meta for out of order Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix windows error Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> * Fix review comments Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Jesus Vazquez <jesus.vazquez@grafana.com> Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com> Signed-off-by: Ganesh Vernekar 15064823+codesome@users.noreply.github.com Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com> Co-authored-by: Dieter Plaetinck <dieter@grafana.com> Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com> Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
2 years ago
if rand.Intn(2) == 0 {
isOOO = true
}
chunkRef = hrw.WriteChunk(seriesRef, mint, maxt, chunk, isOOO, func(cbErr error) {
Write chunks via queue, predicting the refs (#10051) * Write chunks via queue, predicting the refs Our load tests have shown that there is a latency spike in the remote write handler whenever the head chunks need to be written, because chunkDiskMapper.WriteChunk() blocks until the chunks are written to disk. This adds a queue to the chunk disk mapper which makes the WriteChunk() method non-blocking unless the queue is full. Reads can still be served from the queue. Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feeddback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * initialize metrics without .Add(0) Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * change isRunningMtx to normal lock Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * do not re-initialize chunkrefmap Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update metric outside of lock scope Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * add benchmark for adding job to chunk write queue Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * remove unnecessary "success" var Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * gofumpt -extra Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * avoid WithLabelValues call in addJob Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * format comments Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * addressing PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * rename cutExpectRef to cutAndExpectRef Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * use head.Init() instead of .initTime() Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * address PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * PR feedback Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com> Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * update test according to PR feedback Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * better test of truncation with empty files Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> * replace callbackWg -> awaitCb Signed-off-by: Mauro Stettler <mauro.stettler@gmail.com> Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
3 years ago
require.NoError(t, err)
close(awaitCb)
})
<-awaitCb
return
}