`TaskManager#commitAll()` still commits both, active and standby tasks. What 
kind of separation do you mean? And even if we separate both, it seems to be 
orthogonal to my comment.

If we commit all tasks because commit time elapses, we don't need to check for 
user requested commits any longer. Atm, we might iterator over all tasks twice. 
First iteration is checking for user requested commits, and if commit interval 
is passed, we iterate over all tasks again. However, if we commit time passed, 
we commit all tasks anyway and thus can avoid checking for user requested 
commits (ie, we can put `int committed = taskManager.maybeCommitActiveTasks();` 
into the `else` of `if (commitTimeMs >= 0 && lastCommitMs + commitTimeMs < 
now)` ?

Or maybe I miss understood your comment?

[ Full content available at: https://github.com/apache/kafka/pull/5428 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to