Skip to content

Commit 7a81c78

Browse files
alexqylealanprotdsabsayDaniel Sabsaydependabot[bot]
authored
Create guide doc for partition compaction (#6512)
* Purge expired postings cache items due inactivity (#6502) * Purge expired postings cache items due inactivity Signed-off-by: alanprot <alanprot@gmail.com> * Fix comments Signed-off-by: alanprot <alanprot@gmail.com> --------- Signed-off-by: alanprot <alanprot@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Update thanos to 4ba0ba403896 (#6503) * Update thanos to 4ba0ba403896 Signed-off-by: Daniel Sabsay <sabsay@adobe.com> * run go mod vendor Signed-off-by: Daniel Sabsay <sabsay@adobe.com> --------- Signed-off-by: Daniel Sabsay <sabsay@adobe.com> Co-authored-by: Daniel Sabsay <sabsay@adobe.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Bump the actions-dependencies group across 1 directory with 2 updates (#6505) Bumps the actions-dependencies group with 2 updates in the / directory: [actions/upload-artifact](https://github.com/actions/upload-artifact) and [github/codeql-action](https://github.com/github/codeql-action). Updates `actions/upload-artifact` from 4.5.0 to 4.6.0 - [Release notes](https://github.com/actions/upload-artifact/releases) - [Commits](actions/upload-artifact@6f51ac0...65c4c4a) Updates `github/codeql-action` from 3.28.0 to 3.28.1 - [Release notes](https://github.com/github/codeql-action/releases) - [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md) - [Commits](github/codeql-action@48ab28a...b6a472f) --- updated-dependencies: - dependency-name: actions/upload-artifact dependency-type: direct:production update-type: version-update:semver-minor dependency-group: actions-dependencies - dependency-name: github/codeql-action dependency-type: direct:production update-type: version-update:semver-patch dependency-group: actions-dependencies ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * calculate # of concurrency only once at the runner (#6506) Signed-off-by: SungJin1212 <tjdwls1201@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Implement partition compaction planner (#6469) * Implement partition compaction grouper Signed-off-by: Alex Le <leqiyue@amazon.com> * fix comment Signed-off-by: Alex Le <leqiyue@amazon.com> * replace level 1 compaction limits with ingestion replication factor Signed-off-by: Alex Le <leqiyue@amazon.com> * fix doc Signed-off-by: Alex Le <leqiyue@amazon.com> * update compaction_visit_marker_timeout default value Signed-off-by: Alex Le <leqiyue@amazon.com> * update default value for compactor_partition_index_size_limit_in_bytes Signed-off-by: Alex Le <leqiyue@amazon.com> * refactor code Signed-off-by: Alex Le <leqiyue@amazon.com> * address comments and refactor Signed-off-by: Alex Le <leqiyue@amazon.com> * address comment Signed-off-by: Alex Le <leqiyue@amazon.com> * address comment Signed-off-by: Alex Le <leqiyue@amazon.com> * update config name Signed-off-by: Alex Le <leqiyue@amazon.com> * Implement partition compaction planner Signed-off-by: Alex Le <leqiyue@amazon.com> * fix after rebase Signed-off-by: Alex Le <leqiyue@amazon.com> * addressed comments Signed-off-by: Alex Le <leqiyue@amazon.com> * updated doc and refactored metric Signed-off-by: Alex Le <leqiyue@amazon.com> * fix test Signed-off-by: Alex Le <leqiyue@amazon.com> --------- Signed-off-by: Alex Le <leqiyue@amazon.com> * Add max tenant config to tenant federation (#6493) Signed-off-by: SungJin1212 <tjdwls1201@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Add cleaner logic to clean partition compaction blocks and related files (#6507) * Add cleaner logic to clean partition compaction blocks and related files Signed-off-by: Alex Le <leqiyue@amazon.com> * refactored metrics Signed-off-by: Alex Le <leqiyue@amazon.com> * refactor Signed-off-by: Alex Le <leqiyue@amazon.com> * update logs Signed-off-by: Alex Le <leqiyue@amazon.com> --------- Signed-off-by: Alex Le <leqiyue@amazon.com> * Update RELEASE.md (#6511) Maintainers would like an additional week to get the partition compactor changes in before the first release candidate for 1.19. Signed-off-by: Charlie Le <charlie_le@apple.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * update thanos version to 236777732278c64ca01c1c09d726f0f712c87164 (#6514) Signed-off-by: yeya24 <benye@amazon.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Fix race that can cause nil reference when using expanded postings (#6518) Signed-off-by: alanprot <alanprot@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Add more op label values to cortex_query_frontend_queries_total metric (#6519) Signed-off-by: Alex Le <leqiyue@amazon.com> * Allow use of non-dualstack endpoints for S3 blocks storage (#6522) Signed-off-by: Alex Le <leqiyue@amazon.com> * Expose grpc client connect timeout config and default to 5s (#6523) * expose grpc client connect timeout config Signed-off-by: yeya24 <benye@amazon.com> * changelog Signed-off-by: yeya24 <benye@amazon.com> --------- Signed-off-by: yeya24 <benye@amazon.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Hook up partition compaction end to end implementation (#6510) * Implemented partition compaction end to end with custom compaction lifecycle Signed-off-by: Alex Le <leqiyue@amazon.com> * removed unused variable Signed-off-by: Alex Le <leqiyue@amazon.com> * tweak test Signed-off-by: Alex Le <leqiyue@amazon.com> * tweak test Signed-off-by: Alex Le <leqiyue@amazon.com> * refactor according to comments Signed-off-by: Alex Le <leqiyue@amazon.com> * tweak test Signed-off-by: Alex Le <leqiyue@amazon.com> * check context error inside sharded posting Signed-off-by: Alex Le <leqiyue@amazon.com> * fix lint Signed-off-by: Alex Le <leqiyue@amazon.com> * fix integration test for memberlist Signed-off-by: Alex Le <leqiyue@amazon.com> * make compactor initial wait cancellable Signed-off-by: Alex Le <leqiyue@amazon.com> --------- Signed-off-by: Alex Le <leqiyue@amazon.com> * Test for nil on expire expanded postings (#6521) * Test for nil on expire expanded postings Signed-off-by: alanprot <alanprot@gmail.com> * stopping ingester Signed-off-by: alanprot <alanprot@gmail.com> * refactor the test to not timeout Signed-off-by: alanprot <alanprot@gmail.com> --------- Signed-off-by: alanprot <alanprot@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * log when a request starts running in querier (#6525) * log when a request starts running in querier Signed-off-by: Ahmed Hassan <afayekhassan@gmail.com> * log when a request starts running in querier for frontend processor Signed-off-by: Ahmed Hassan <afayekhassan@gmail.com> --------- Signed-off-by: Ahmed Hassan <afayekhassan@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Update build image according to 03a8f8c (#6508) Signed-off-by: Friedrich Gonzalez <friedrichg@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Deprecate -blocks-storage.tsdb.wal-compression-enabled flag Signed-off-by: SungJin1212 <tjdwls1201@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Fix test (#6537) Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Mark 1.19 release in progress https://github.com/cortexproject/cortex/blob/master/RELEASE.md#show-that-a-release-is-in-progress Signed-off-by: Charlie Le <charlie_le@apple.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Prepare 1.19.0-rc.0 Signed-off-by: Charlie Le <charlie_le@apple.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Revert "Prepare 1.19.0-rc.0" Signed-off-by: Alex Le <leqiyue@amazon.com> * Fixed blocksGroupWithPartition unable to reuse functions from blocksGroup (#6547) * Fixed blocksGroupWithPartition unable to reuse functions from blocksGroup Signed-off-by: Alex Le <leqiyue@amazon.com> * update tests Signed-off-by: Alex Le <leqiyue@amazon.com> --------- Signed-off-by: Alex Le <leqiyue@amazon.com> * Remove TransferChunks gRPC method (#6543) Signed-off-by: SungJin1212 <tjdwls1201@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Uupdate Ppromqlsmith (#6557) Signed-off-by: alanprot <alanprot@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Query Partial Data (#6526) * Create partial_data Signed-off-by: Justin Jung <jungjust@amazon.com> * Fix lazyquery so that warning message is returned Signed-off-by: Justin Jung <jungjust@amazon.com> * Add QueryPartialData limit Signed-off-by: Justin Jung <jungjust@amazon.com> * Fix broken mock Signed-off-by: Justin Jung <jungjust@amazon.com> * Make response with warnings to be not cached Signed-off-by: Justin Jung <jungjust@amazon.com> * Updated streamingSelect in distributor_queryable Signed-off-by: Justin Jung <jungjust@amazon.com> * Update query.go Signed-off-by: Justin Jung <jungjust@amazon.com> * Update replication_set Signed-off-by: Justin Jung <jungjust@amazon.com> * Lint Signed-off-by: Justin Jung <jungjust@amazon.com> * Lint again Signed-off-by: Justin Jung <jungjust@amazon.com> * Generated doc Signed-off-by: Justin Jung <jungjust@amazon.com> * Changelog Signed-off-by: Justin Jung <jungjust@amazon.com> * Update config description Signed-off-by: Justin Jung <jungjust@amazon.com> * Do not remove warnings from seriesSet Signed-off-by: Justin Jung <jungjust@amazon.com> * Avoid cache only if the warning message contains partial data error Signed-off-by: Justin Jung <jungjust@amazon.com> * Remove context usage for partial data Signed-off-by: Justin Jung <jungjust@amazon.com> * Refactor how partial data info is passed + apply to series and label methods as well Signed-off-by: Justin Jung <jungjust@amazon.com> * Lint + fix tests Signed-off-by: Justin Jung <jungjust@amazon.com> * Fix build Signed-off-by: Justin Jung <jungjust@amazon.com> * Create separate config for ruler partial data Signed-off-by: Justin Jung <jungjust@amazon.com> * Genereta doc Signed-off-by: Justin Jung <jungjust@amazon.com> * Add more tests Signed-off-by: Justin Jung <jungjust@amazon.com> * Change error Signed-off-by: Justin Jung <jungjust@amazon.com> * Fix test Signed-off-by: Justin Jung <jungjust@amazon.com> * Update changelog Signed-off-by: Justin Jung <jungjust@amazon.com> * Update changelog Signed-off-by: Justin Jung <jungjust@amazon.com> * Nit Signed-off-by: Justin Jung <jungjust@amazon.com> * Nit Signed-off-by: Justin Jung <jungjust@amazon.com> --------- Signed-off-by: Justin Jung <jungjust@amazon.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Add timeout for dynamodb ring kv (#6544) * add dynamodb kv with timeout enforced Signed-off-by: yeya24 <benye@amazon.com> * add tests Signed-off-by: yeya24 <benye@amazon.com> * docs Signed-off-by: Ben Ye <benye@amazon.com> * update changelog Signed-off-by: Ben Ye <benye@amazon.com> --------- Signed-off-by: yeya24 <benye@amazon.com> Signed-off-by: Ben Ye <benye@amazon.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Bump the actions-dependencies group across 1 directory with 2 updates (#6564) Bumps the actions-dependencies group with 2 updates in the / directory: [github/codeql-action](https://github.com/github/codeql-action) and [actions/setup-go](https://github.com/actions/setup-go). Updates `github/codeql-action` from 3.28.1 to 3.28.7 - [Release notes](https://github.com/github/codeql-action/releases) - [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md) - [Commits](github/codeql-action@b6a472f...6e54559) Updates `actions/setup-go` from 5.2.0 to 5.3.0 - [Release notes](https://github.com/actions/setup-go/releases) - [Commits](actions/setup-go@3041bf5...f111f33) --- updated-dependencies: - dependency-name: github/codeql-action dependency-type: direct:production update-type: version-update:semver-patch dependency-group: actions-dependencies - dependency-name: actions/setup-go dependency-type: direct:production update-type: version-update:semver-minor dependency-group: actions-dependencies ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Fix: expanded postings can cache wrong data when queries are issued "in the future" (#6562) * improve fuzz test for expanded postings cache Signed-off-by: alanprot <alanprot@gmail.com> * create more tests on the expanded postings cache Signed-off-by: alanprot <alanprot@gmail.com> * adding get series call on the test Signed-off-by: alanprot <alanprot@gmail.com> * no use CachedBlockChunkQuerier when query time range is completely after the last sample added in the head Signed-off-by: alanprot <alanprot@gmail.com> * adding comments Signed-off-by: alanprot <alanprot@gmail.com> * increase the number of fuzz test from 100 to 300 Signed-off-by: alanprot <alanprot@gmail.com> * add get series fuzzy testing Signed-off-by: alanprot <alanprot@gmail.com> --------- Signed-off-by: alanprot <alanprot@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Extend ShuffleSharding on READONLY ingesters (#6517) * Filter readOnly ingesters when sharding Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> * Extend shard on READONLY Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> * Remove old code Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> * Fix test Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> * update changelog Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> --------- Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * Create guide doc for partition compaction Signed-off-by: Alex Le <leqiyue@amazon.com> * Update docs/guides/partitioning-compactor.md Co-authored-by: Charlie Le <charlie_le@apple.com> Signed-off-by: Alex Le <emoc1989@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> * updated doc Signed-off-by: Alex Le <leqiyue@amazon.com> * clean white space Signed-off-by: Alex Le <leqiyue@amazon.com> --------- Signed-off-by: alanprot <alanprot@gmail.com> Signed-off-by: Alex Le <leqiyue@amazon.com> Signed-off-by: Daniel Sabsay <sabsay@adobe.com> Signed-off-by: dependabot[bot] <support@github.com> Signed-off-by: SungJin1212 <tjdwls1201@gmail.com> Signed-off-by: Charlie Le <charlie_le@apple.com> Signed-off-by: yeya24 <benye@amazon.com> Signed-off-by: Ahmed Hassan <afayekhassan@gmail.com> Signed-off-by: Friedrich Gonzalez <friedrichg@gmail.com> Signed-off-by: Daniel Deluiggi <ddeluigg@amazon.com> Signed-off-by: Justin Jung <jungjust@amazon.com> Signed-off-by: Ben Ye <benye@amazon.com> Signed-off-by: Alex Le <emoc1989@gmail.com> Co-authored-by: Alan Protasio <approtas@amazon.com> Co-authored-by: Daniel Sabsay <danielrsabsay@gmail.com> Co-authored-by: Daniel Sabsay <sabsay@adobe.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: SungJin1212 <tjdwls1201@gmail.com> Co-authored-by: Charlie Le <charlie_le@apple.com> Co-authored-by: Ben Ye <benye@amazon.com> Co-authored-by: Sam McBroom <86423878+sam-mcbr@users.noreply.github.com> Co-authored-by: Ahmed Hassan <57634502+afhassan@users.noreply.github.com> Co-authored-by: Friedrich Gonzalez <1517449+friedrichg@users.noreply.github.com> Co-authored-by: Daniel Blando <daniel@blando.com.br> Co-authored-by: Justin Jung <jungjust@amazon.com>
1 parent 6078955 commit 7a81c78

File tree

1 file changed

+50
-0
lines changed

1 file changed

+50
-0
lines changed

docs/guides/partitioning-compactor.md

Lines changed: 50 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,50 @@
1+
---
2+
title: "Use Partition Compaction in Cortex"
3+
linkTitle: "Partition Compaction"
4+
weight: 10
5+
slug: partition-compaction
6+
---
7+
8+
## Context
9+
10+
Compactor is bounded by maximum 64GB of index file size. If compaction failed due to exceeding index file size limit, partition compaction can be enabled to allow compactor compacting into multiple blocks that have index file size stays within limit.
11+
12+
## Enable Partition Compaction
13+
14+
In order to enable partition compaction, the following flag needs to be set:
15+
16+
```
17+
-compactor.sharding-enabled=true # Enable sharding tenants across multiple compactor instances. This is required to enable partition compaction
18+
-compactor.sharding-strategy=shuffle-sharding # Use Shuffle Sharding as sharding strategy. This is required to enable partition compaction
19+
-compactor.compaction-strategy=partitioning # Use Partition Compaction as compaction strategy. To turn if off, set it to `default`
20+
```
21+
22+
### Migration
23+
24+
There is no special migration process needed to enable partition compaction. End user could enable it by setting the above configurations all at once.
25+
26+
Enabling partition compaction would group previously compacted blocks (only those have time range smaller than the largest configured compaction time ranges) with uncompacted blocks and generate new compaction plans. This would group blocks having duplicated series together and those series would be deduped after compaction.
27+
28+
Disabling partition compaction after enabled it does not need migration either. After disabling partition compaction, compactor would group partitioned result blocks together and compact them into one block.
29+
30+
## Configure Partition Compaction
31+
32+
By default, partition compaction utilizes the following configurations and their values:
33+
34+
```
35+
-compactor.partition-index-size-bytes=68719476736 # 64GB
36+
-compactor.partition-series-count=0 # no limit
37+
```
38+
39+
The default value should start partitioning result blocks when sum of index files size of parent blocks exceeds 64GB. End user could also change those two configurations. Partition compaction would always calculate partition count based on both configuration and pick the one with higher partition count.
40+
41+
Both configurations support to be set per tenant.
42+
43+
Note: `compactor.partition-series-count` is using sum of series count of all parent blocks. If parent blocks were not deduped, the result block could have fewer series than the configuration value.
44+
45+
## Useful Metrics
46+
47+
- `cortex_compactor_group_partition_count`: can be used to keep track of how many partitions being compacted for each time range.
48+
- `cortex_compactor_group_compactions_not_planned_total`: can be used to alarm any compaction was failed to be planned due to error.
49+
- `cortex_compact_group_compaction_duration_seconds`: can be used to monitor compaction duration of each time range compactions.
50+
- `cortex_compactor_oldest_partition_offset`: can be used to monitor when was the oldest compaction that is still not completed.

0 commit comments

Comments
 (0)