Making the number of CPUs used for WAL replay configurable #4445
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
treid314
reviewed
Mar 9, 2023
There was a problem hiding this comment.
Style suggestions mostly, I think what's here makes sense. When Prometheus got updated it looks like they moved the tsdb.GenerateTestHistogram function to tsdbutil. which is causing the unit test and linting issues. Not sure if we have a mainline PR open to update Prometheus yet but I don't think we should be doing that here.
|
The referenced pull request #1040 seems unrelated to this change? |
56quarters
reviewed
Mar 9, 2023
duricanikolic
force-pushed
the
yuri/wal-replay-optimization
branch
from
d8bedb3
to
1be6b7b
Compare
pracucci
reviewed
Mar 10, 2023
There was a problem hiding this comment.
Some notes:
- List the deprecated config option under
docs/sources/mimir/operators-guide/configure/about-versioning.md - In the vendored Prometheus they've added
lookback_deltaparameter to the instant and range query API. To be 100% Prometheus compatible we should support it too. This is not expected to be done in this PR, but please open an issue in the Mimir repo about it and kick off the discussion whether we should add support for it (I think so) - I checked other vendored changes and LGTM
pracucci
reviewed
Mar 10, 2023
duricanikolic
force-pushed
the
yuri/wal-replay-optimization
branch
from
e70fa6a
to
789fbf0
Compare
pracucci
reviewed
Mar 10, 2023
pracucci
approved these changes
Mar 10, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What this PR does
The version of
prometheusis upgraded (viamimir-prometheus) to the one corresponding to this commit, which introducedWALReplayConcurrencyas an option on tsdbOptionsandHeadOptions. This option represents the maximal number of CPUs used during WAL replay on TSDB opening. If not set to a positive number, all available CPUs are used.WALReplayConcurrencyhas been added as a tsdb option onmimirside as well, and it is configurable via the newly added advanced configuration parameter-blocks-storage.tsdb.wal-replay-concurrency, whose default value is 0. The latter is supposed to override the-blocks-storage.tsdb.max-tsdb-opening-concurrency-on-startup, which is now deprecated, and will be removed fromMimir 2.10. In order to guarantee backwards compatibility, the following logic is used:-blocks-storage.tsdb.wal-replay-concurrencyis set to a positive number, the deprecated-blocks-storage.tsdb.max-tsdb-opening-concurrency-on-startupis ignored.-blocks-storage.tsdb.max-tsdb-opening-concurrency-on-startupis used and the behavior is identical to the one before this PR.Setting
-blocks-storage.tsdb.wal-replay-concurrencyto a positive value imposes the maximal number of concurrent processes to be used on ingesters startup. The heuristic used is the following:10tenants, opening of TSDBs is done by-blocks-storage.tsdb.wal-replay-concurrencyconcurrent processes, each of them using only 1 process for WAL replay (i.e.,prometheusoptionWALReplayConcurrencyis set to1).-blocks-storage.tsdb.wal-replay-concurrencyconcurrent processes for WAL replay ((i.e.,prometheusoptionWALReplayConcurrencyis set to-blocks-storage.tsdb.wal-replay-concurrency).Which issue(s) this PR fixes or relates to
Part of https://github.com/grafana/mimir-squad/issues/1040
Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX]