ingest: add long polling to TestConcurrentFetchers #9971
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
Some of the tests assume that we'd fetch all produced records in one go. The Kafka protocol doesn't guarantee that and sometimes kfake also returns less than that. This means that sometimes the tests would finish early before we've fetched all records. This PR introduced `logPollFetches` which loops over `fetchers.PollFetches` multiple times until a timeout or until we fetch the expected number of records. I set the timeout to 2s because of the MinBytesMaxWait of 1s in tests ([code](https://github.com/grafana/mimir/blob/96c92c99ef5599d4507c2aa9f6d7ddf42d5beec2/pkg/storage/ingest/fetcher_test.go#L1373)): * fetcher receives 1 record (1st request) * fetcher sends the `fetchResult` to `fetchWant.result` * fetcher receives 2 records (2nd request) * fetcher can't send the result to `fetchWant.result` because `start()` isn't receiving on the channel; `start()` is trying to send to `orderedFetches` * fetcher start another (3rd) attempt; the attempt takes 1s (`MaxWaitMillis` in the Fetch request) * client calls `PollFetches`, reads from orderedFetches; receives the first `Fetches` with 1 records; `start()` goes back to waiting on `fetchWant.result` * client calls `PollFetches`, `start()` is still waiting on `fetchWant.result` * fetcher receives the response to 3rd request with no records; sends the buffered 2 records from 2nd request * client receives 2 records Another solution to waiting for 2s is to make the merging of results in `start()` for example. Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
dimitarvdimitrov
force-pushed
the
dimitar/ingest/add-long-polling-to-TestConcurrentFetchers
branch
from
4303e05 to
ce89479
Compare
Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
Signed-off-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
dimitarvdimitrov
deleted the
dimitar/ingest/add-long-polling-to-TestConcurrentFetchers
branch
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
Some of the tests assume that we'd fetch all produced records in one go. The Kafka protocol doesn't guarantee that and sometimes kfake also returns less than that. This means that sometimes the tests would finish early before we've fetched all records.
This PR introduced
logPollFetcheswhich loops overfetchers.PollFetchesmultiple times until a timeout or until we fetch the expected number of records.I also changed all places which had this assumption and all places which already did some form of log polling.
2s timeout
I set the timeout to 2s because of the MinBytesMaxWait of 1s in tests (code):
fetchResulttofetchWant.resultfetchWant.resultbecausestart()isn't receiving on the channel;start()is trying to send toorderedFetchesMaxWaitMillisin the Fetch request)PollFetches, reads from orderedFetches; receives the firstFetcheswith 1 records;start()goes back to waiting onfetchWant.resultPollFetches,start()is still waiting onfetchWant.resultAnother solution to waiting for 2s is to make the merging of results in
start()for example.Which issue(s) this PR fixes or relates to
Fixes #9970
Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX].about-versioning.mdupdated with experimental features.