github-actions[bot] commented on code in PR #64813:
URL: https://github.com/apache/doris/pull/64813#discussion_r3473926007
##########
fe/fe-core/src/main/java/org/apache/doris/cloud/CloudWarmUpJob.java:
##########
@@ -990,13 +995,17 @@ private void runEventDrivenJob() throws Exception {
hasTableFilter() ? getCurrentTableIdNames().size() :
"all");
TWarmUpTabletsResponse response =
entry.getValue().warmUpTablets(request);
if (response.getStatus().getStatusCode() != TStatusCode.OK) {
+ hasError = true;
if (!response.getStatus().getErrorMsgs().isEmpty()) {
errMsg = response.getStatus().getErrorMsgs().get(0);
}
LOG.warn("send warm up request failed. job_id={},
event={}, err={}",
jobId, syncEvent, errMsg);
}
}
Review Comment:
This clears the stale error only in memory for event-driven jobs. After
`runEventDrivenJob()` returns, the event-driven branch in `runRunningJob()`
returns immediately without writing a `logModifyCloudWarmUpJob`, while the
batched success path below clears the same field and then journals the job
before returning. Since `errMsg` is part of the serialized `CloudWarmUpJob`, a
concrete sequence is: a prior event-driven failure is checkpointed with a
non-empty error, a later successful retry clears this field only in memory,
then FE restarts before the next checkpoint and replays the old image with the
stale error, so `SHOW WARM UP JOB` shows the old failure again. Please persist
the reset when it changes, for example by logging the job after a successful
all-BE event-driven `SET_JOB` clears a non-empty message, and add a test that
verifies the journal/replay behavior.
--
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]
