Operating System: All
Platform: All

We are seeing noticeably more sporadic failures in unit tests that cause the autorelease build to fail pretty much every time. My current theory is that the timings of Karaf 4 being different from Karaf 3 cause that to happen, but I'm honestly not 100% sure about that.

In any event, we need to have ways to get past that, and the current plan has 4 parallel prongs:

1.) Make a new autorelease-carbon job that runs with -DskipTests so we can produce artifacts that we can run CSIT against and do manual testing with.

2.) Make a new autorelease-carbon job that runs with -fn so that it can provide us with a fast way to identify unit test failures and report them out to projects to get them fixed or ignore them going forward.

3.) Add a parameter to the current autorelease-carbon job that takes an ExcludesFile of tests to ignore when building so as we find sporadically failing tests, we can ignore them.

4.) Actually get unit tests that are failing fixed.

There's a long thread starting here:
https://lists.opendaylight.org/pipermail/release/2017-April/010290.html

I'm going to create new bugs for 1-3 above so we can track them independently and this will depend on them.

Issue 1.) 2.) and 3.) have been resolved and fixed in different bugs. For 4.) we need to open a separate bug with the project to get it resolved. Do we still need to keep this bug as blocker ?

I've changed the requirement for 4 to simply having the a way to disable SingleFeatureTests in autorelease, which is tracked by BUG-8365 now.

Can we reduce the priority or close the bug, since all the four issue have been addressed in separate bugs ?

Sounds like there's nothing left to do here. Please reopen this with more details if anyone disagrees.