auto-test intermittently doesn't trigger in HDFS project
--------------------------------------------------------
Key: HDFS-1830
URL: https://issues.apache.org/jira/browse/HDFS-1830
Project: Hadoop HDFS
Issue Type: Test
Components: test
Reporter: Matt Foley
For HDFS-1295, the Hudson QA test would not run on the 02/Apr/11 01:13
submission, IBR_shortcut_v3atrunk.patch [ 12475272 ], until Nicholas (szetszwo)
triggered it manually for me.
Now, with the 11/Apr/11 06:59 submission of IBR_shortcut_v4atrunk.patch [
12475977 ],
and with the re-submission at 11/Apr/11 20:26 of IBR_shortcut_v4atrunk.patch [
12476054 ],
and finally another at 11/Apr/11 22:26 of IBR_shortcut_v4atrunk.patch [
12476063 ], with status first set to "Resume Work"/"In Progress",
it is again failing to pick up the existence of the new patch candidate.
Finally, Konstantin (shv) started it manually.
I have also observed the same problem on HDFS-1070.
The 08/Apr/11 17:21 submission of trunkLocalNameImage7.patch [ 12475829 ] did
not get tested until I asked Nicholas to trigger it manually.
And the 11/Apr/11 21:21 submission of trunkLocalNameImage8.patch [ 12476058 ]
has been sitting for 1.5 hours, and hasn't been picked up, although it is my
understanding that the need should be polled every 10 minutes.
I've also observed at
https://builds.apache.org/hudson/job/PreCommit-HDFS-Build/ that it often has up
to three very long-running jobs in queue. Is it perhaps the case that patches
posted when the queue is full do not get picked up after the queue is emptied?
Unquestionably it often DOES work, but equally clearly it doesn't always when
it should.
When I originally posted this yesterday, PreCommit-HDFS-Build had been entirely
idle for about 20 minutes, until build 346 was started manually.
All times are in GMT, as in Jira.
Thank you for your help. If the Jira event log indicates I'm not doing
something right, please inform.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
