lhotari commented on PR #23903:
URL: https://github.com/apache/pulsar/pull/23903#issuecomment-2618373779

   For preventing future regressions like #23900, it would be necessary to add 
specific tests for that purpose, to ensure that replay queue reads don't cause 
cache misses in a tailing read scenario. The regression is specific to tailing 
reads.
   
   In the current broker cache implementation, catchup reads will result in 
cache misses for replay queue reads. This behavior is not a regression. This 
has a performance impact at least on Key_Shared subscriptions where it's common 
that entries get added to the replay queue. It's a future improvement that I'm 
planning to address that there would be caching also for that type of scenario.
   
   Since we don't have currently sufficient test coverage for various 
scenarios, regressions like this can happen.
   The goal of recent test framework improvements such as #23875 and #23892 are 
to make it easier to write tests also for this type of scenarios.


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to