gianm commented on code in PR #18936:
URL: https://github.com/apache/druid/pull/18936#discussion_r2722066488
##########
indexing-service/src/main/java/org/apache/druid/indexing/seekablestream/supervisor/SeekableStreamSupervisor.java:
##########
@@ -3461,6 +3461,7 @@ void maybeScaleDuringTaskRollover()
if (rolloverTaskCount > 0) {
log.info("Autoscaler recommends scaling down to [%d] tasks during
rollover", rolloverTaskCount);
changeTaskCountInIOConfig(rolloverTaskCount);
+ autoScalerConfig.setTaskCountStart(rolloverTaskCount);
Review Comment:
I don't quite understand what you mean by "it doesn't work on the task
level". The autoscaler config is a supervisor-only concept so it should only
need to be persisted with the supervisor. I'm asking about this because I
thought the conclusion in #18745 did make sense, and I feel the current PR
deviates from that approach. In particular, in #18745 we arrived at an approach
where `taskCount` is always the "current" task count and is updated by the
autoscaler as the task count changes. `taskCountStart` is provided by the user
only, and is optional. If present, it overwrites `taskCount` on supervisor
submit (but not stop/start or failover).
--
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]
