[
https://issues.apache.org/jira/browse/HBASE-21319?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16651918#comment-16651918
]
Sean Busbey commented on HBASE-21319:
-------------------------------------
we could do --fail-at-end and have things work normally. I don't think we can
do --fail-never without also adding some logic to parse the surefire results to
see if we should fail things.
a couple of other ways we could improve this:
* move off of the first / second group stuff so that within one module we don't
fail to run the long tests because the short tests fail (I get why we do this
for developers, but things are different in CI IMHO)
* we probably could configure our yetus personality to run individual modules
for unit tests instead of running a top level build as it currently does.
if we did both of these we'd get a full run with failures accounted for even in
the face of some things failing.
> The nightly jobs should run 'mvn test' with '-fn'
> -------------------------------------------------
>
> Key: HBASE-21319
> URL: https://issues.apache.org/jira/browse/HBASE-21319
> Project: HBase
> Issue Type: Bug
> Components: test
> Reporter: Duo Zhang
> Priority: Major
>
> For now, it will fail in the middle if there are failures. I think for a
> nightly job we should run all the UTs.
> The nightly pipeline is complicated and uses yetus so I'm not sure how to fix
> it...
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
