Tarun-kishore commented on PR #20118:
URL: https://github.com/apache/pinot/pull/20118#issuecomment-6572198172
Thanks. You're right on TTL: the sweep only runs when a consuming segment
starts, and the threshold is anchored to the largest consumed comparison value,
so it can't bring a paused table under the cap. I corrected the javadoc.
Retention deletes, a restart, or raising the cap are the ways out.
On partial responses, which the model didn't cover: that was a real gap. The
metadata reader drops failed servers silently, so a missed response could
resume a PK-paused table. A check is now only "within quota" when every server
hosting the table responded. A breach seen in a partial response still counts,
since missing servers can only hide keys. Tests added for both cases.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
