cadonna commented on PR #15640: URL: https://github.com/apache/kafka/pull/15640#issuecomment-2117012544
> High level comment, just to clarify and make sure it's something we are considering and will cover with the follow-up PRs for timeouts. Here we're introducing a component to ensures that app events are expired only after having one chance, but that's only at the app thread level, and not for all events, but only for unsubscribe, and poll. Thing is that events can also be expired indirectly when a request is expired (so playing against this changes). So even if the `processBackgroundEvents` introduced here gives an expired event a change to run one, that may actually not happen (because the underlying request expires). I expect that side needed to make this whole intention work in practice will be a follow-up PR, am I right? > > Also note, almost none of our integration test cover the poll(ZERO) case (helper funcs [here](https://github.com/apache/kafka/blob/056d232f4e28bf8e67e00f8ed2c103fdb0f3b78e/core/src/test/scala/unit/kafka/utils/TestUtils.scala#L892-L910), used by most of the test, poll with timeout > 0). That's probably why we did not find the expiration issues we have with poll(0) before. I guess that after addressing the timeout/expiration (this PR and follow-up), we should be able to add some. @lianetm Could you elaborate on this a bit. I cannot completely follow. I followed the `SyncEvent`. The event is processed in the `ApplicationEventHandler` and a request is added to the commit request manager. Then the commit request manager is polled, the requests are added to the network client and the the network client is polled. As far as I can see the `SyncEvent` is given a chance to complete. I did also not understand the connection to `processBackgroundEvents`. That is for events that originate from the background thread. How does that influence if an application event was given a chance to complete? What do you mean with the "underlying requests expires"? Do you mean it exceeds the request timeout? Does the legacy consumer ensure that a request completed when timeout is set to `0`? Sorry for all these questions, but I would like to understand. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
