github-actions[bot] commented on code in PR #65049:
URL: https://github.com/apache/doris/pull/65049#discussion_r3502806750
##########
fe/fe-core/src/main/java/org/apache/doris/cloud/load/CloudRoutineLoadManager.java:
##########
@@ -55,6 +55,7 @@ protected List<Long> getAvailableBackendIds(long jobId)
throws LoadException {
.getBackendsByClusterName(routineLoadJob.getCloudCluster())
.stream()
.filter(Backend::isAlive)
Review Comment:
Cloud routine load still only starts from `Backend::isAlive` here before
applying the new decommissioning filter. That fixes the draining-BE case, but
it leaves this cloud selector weaker than the local routine-load path:
`Backend.isLoadAvailable()` also rejects `setLoadDisabled(true)` and shutdown
BEs, and `RoutineLoadTaskScheduler.allocateTaskToBe()` assigns the returned id
directly before submitting the `TRoutineLoadTask`. A cloud job can therefore
still schedule new routine-load work to a BE where load has been disabled.
Please use the same load-available predicate/policy semantics here, and extend
the cloud routine-load test with a load-disabled alive backend.
--
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]
