ingest: use single configuration for fetch concurrency #10156
Merged
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
dimitarvdimitrov
commented
Dec 6, 2024
| @@ -71,7 +70,6 @@ | |||
| * [ENHANCEMENT] Compactor: refresh deletion marks when updating the bucket index concurrently. This speeds up updating the bucket index by up to 16 times when there is a lot of blocks churn (thousands of blocks churning every cleanup cycle). #9881 | |||
| * [ENHANCEMENT] PromQL: make `sort_by_label` stable. #9879 | |||
| * [ENHANCEMENT] Distributor: Initialize ha_tracker cache before ha_tracker and distributor reach running state and begin serving writes. #9826 #9976 | |||
| * [ENHANCEMENT] Ingester: `-ingest-storage.kafka.max-buffered-bytes` to limit the memory for buffered records when using concurrent fetching. #9892 | |||
There was a problem hiding this comment.
i've removed this item because they're not changes compared to the last released version because concurrent fetching and ingestion haven't been released yet
dimitarvdimitrov
commented
Dec 6, 2024
| @@ -16,7 +16,6 @@ | |||
| * [CHANGE] Distributor: Drop experimental `-distributor.direct-otlp-translation-enabled` flag, since direct OTLP translation is well tested at this point. #9647 | |||
| * [CHANGE] Ingester: Change `-initial-delay` for circuit breakers to begin when the first request is received, rather than at breaker activation. #9842 | |||
| * [CHANGE] Query-frontend: apply query pruning before query sharding instead of after. #9913 | |||
| * [CHANGE] Ingester: remove experimental flags `-ingest-storage.kafka.ongoing-records-per-fetch` and `-ingest-storage.kafka.startup-records-per-fetch`. They are removed in favour of `-ingest-storage.kafka.max-buffered-bytes`. #9906 | |||
There was a problem hiding this comment.
i've removed this item because they're not changes compared to the last released version because concurrent fetching and ingestion haven't been released yet
This replaces the two previous flags with one single one. ``` # old flags -ingest-storage.kafka.startup-fetch-concurrency -ingest-storage.kafka.ongoing-fetch-concurrency # new flag -ingest-storage.kafka.fetch-concurrency-max ``` At GL we've been running with a single value for both. The reason for having two flags in the first place is because we couldn't find a balance between throughput and latency. This was largely fixed with the change to not issue fetch requests for beyond the HWM (9892). So now we no longer need the complexity of being able to change concurrency settings Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
60e42ef to
fb1082f
Compare
pracucci
approved these changes
Dec 9, 2024
|
The CHANGELOG has just been cut to prepare for the next release. Please rebase |
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
This replaces the two previous flags with one single one.
At GL we've been running with a single value for both. The reason for having two flags in the first place is because we couldn't find a balance between throughput and latency. This was largely fixed with the change to not issue fetch requests for beyond the HWM (#9892). So now we no longer need the complexity of being able to change concurrency settings
Which issue(s) this PR fixes or relates to
Fixes #
Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX].about-versioning.mdupdated with experimental features.