Closed vaikas closed 6 years ago
0.1.1 candidate
@vaikas-google @kibbles-n-bytes Are we still seeing hangs? Does this need to be worked?
I agree we still need the timeout added to the tests
Yes; I added a timeout to Jenkins as a result of this issue to abort the entire build, but I believe we should have timeouts per test.
For integration and e2e, we pass the t *testing.T
into the framework, so maybe setting a timer goroutine on t.Fatalf would work. I think the ones involving the framework are probably most likely to take long enough to need a timeout.
There doesn't seem to be a built-in way to add a timeout to a test. We'd have to write a timeoutFailer
and added it to all the tests.
Don't see anyway to do this in a manner that cleanly moves on to the next test. It would require plumbing through the stopch through every test and then checking if it's time to stop.
The other option is to accept that it's timed out and panic
and stop the whole process.
prow has timeouts per job so this should be solved by having moved most of the test to prow. /close
Due to a bug in the code, our tests were hanging forever and needed a manual abort. https://service-catalog-jenkins.appspot.com/job/service-catalog-PR-testing2/2398/
It was basically waiting for a ServiceClass to be ready looked like. If you revert this commit it will fail: https://github.com/kubernetes-incubator/service-catalog/pull/1319/commits/87ba116213aabee6eeecb738b85e904ca59e2e34