KAFKA-7484: fix suppression integration tests - #5748
Conversation
|
@mjsax Do you think this is likely the fix for the recent test failures? It seems to work for me, but the absence of failure is no proof... |
|
I assume it will work -- but frankly not 100% sure. Don't know exactly, how the broker advances it's mocked time internally. |
|
I also considered setting the broker's log retention to max_int hours, but I figured we should make the minimal change that we think will fix the problem (otherwise, we'll never learn what the actual fix was) |
|
For Integration tests, the Embedded broker will not actively advance its |
guozhangwang
left a comment
There was a problem hiding this comment.
The change lgtm, waiting for jenkins to pass
|
@guozhangwang @mjsax : the tests have passed. |
Reviewers: Guozhang Wang <guozhang@confluent.io>, Matthias J. Sax <matthias@confluent.io>
|
Merged to |
|
Thank you, @mjsax ! |
Reviewers: Guozhang Wang <guozhang@confluent.io>, Matthias J. Sax <matthias@confluent.io>
The test data is based at time 0 (aka 1970), but the broker was started at time
now.This caused the log cleaner to delete the test data if the tests took slightly too long, despite its generous default configuration.
Committer Checklist (excluded from commit message)