New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
AbstractTestNativeGeneralQueries.testSystemTables is flaky #22129
Comments
A Velox problem?
|
It looks like a real problem in Velox / Prestissimo I managed to capture a following sequence of events:
Full stack trace:
|
In Presto the buffer is not deleted on An alternative would be to try to abort pending requests, but it might be more difficult due to inherent "raciness" of these operations. Also it is unclear if there are other sequences of requests possible that can lead to a query failure. @mbasmanova @xiaoxmeng Thoughts? |
I see. It would be nice to update documentation in https://prestodb.io/docs/current/develop/worker-protocol.html to clarify that. Let me check facebookincubator/velox#9207 |
I'm seeing this issue in production from time to time as well: 20240324_170232_00988_y57ss |
Noted, thanks! |
This is a known issue that happens ~10 times a day with failure message |
Summary: Sometimes requests can arrive out of order. When a `getData` arrives after a buffer is closed it may cause an unwanted failure. Fixes prestodb/presto#22129 Pull Request resolved: facebookincubator#9207 Reviewed By: mbasmanova Differential Revision: D55591528 Pulled By: arhimondr fbshipit-source-id: db493477e236b6c25ea39887a3bab5273558455b
TestPrestoNativeGeneralQueriesThrift>AbstractTestNativeGeneralQueries.testSystemTables:1413->AbstractTestQueryFramework.assertQuery:154 Execution of 'actual' query failed: SELECT * FROM (SELECT DISTINCT regionkey FROM tmp_presto_30a13f6f35074dde851d6931c829c9fc) t INNER JOIN (SELECT regionkey FROM "tmp_presto_30a13f6f35074dde851d6931c829c9fc$partitions") p ON t.regionkey = p.regionkey
We've seen a lot of failures in TestPrestoNativeGeneralQueriesThrift. It's possible the whole class (or the superclass?) is flaky rather than individual test methods.
The text was updated successfully, but these errors were encountered: