Github user srowen commented on the pull request:
https://github.com/apache/spark/pull/2108#issuecomment-54893384
I also was unable to reproduce it _in Maven_ when only `SparkSubmitSuite`
was run. This does seem consistent with the failure being caused by port
conflict but I still can't guess why it doesn't happen in SBT.
I found that setting `<forkMode>never</forkMode>` in the `scalatest` plugin
causes this stack trace to show up immediately in the test output too, which is
at least handy for reproducing it.
Another interesting data point is that if I hack the code temporarily to
choose a random UI port, it's all fine.
Choosing ports randomly might be a good idea in the longer-term anyway to
facilitate parallel tests, and I believe is mentioned in another PR.
I suppose it's a potential solution here too, but would be good to
understand why it doesn't happen with SBT. Neither is parallel. Maybe some
difference in the test lifecycle allows the previous web servers to shut down
more reliably before the next ones start? I'm only guessing at this stage.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
