Skip to content

Conversation

@vanzin
Copy link
Contributor

@vanzin vanzin commented Apr 17, 2015

My previous fix (force a selector wakeup) didn't seem to work since
I ran into the hang again. So change the code a bit to be more
explicit about the condition when the selector thread should exit.

My previous fix (force a selector wakeup) didn't seem to work since
I ran into the hang again. So change the code a bit to be more
explicit about the condition when the selector thread should exit.
@vanzin
Copy link
Contributor Author

vanzin commented Apr 17, 2015

I ran the test that hanged a couple of times for me in a loop with this change, and didn't see any hangs. Still, with these things, it's hard to know.

@SparkQA
Copy link

SparkQA commented Apr 18, 2015

Test build #30504 has finished for PR 5566 at commit ddb2c03.

  • This patch passes all tests.
  • This patch merges cleanly.
  • This patch adds no public classes.
  • This patch does not change any dependencies.

@andrewor14
Copy link
Contributor

For those following, this is a hot fix for #5335. @srowen

@srowen
Copy link
Member

srowen commented Apr 18, 2015

I see, did the call to wakeup() create a new problem? I thought it was just trying to solve a hang, but, that itdidn't do the trick. In any event seems fine as an addendum to SPARK-4194.

@asfgit asfgit closed this in 327ebf0 Apr 18, 2015
@vanzin
Copy link
Contributor Author

vanzin commented Apr 20, 2015

No, the change in the previous commit didn't cause a problem (so technically this isn't a hotfix?), but it didn't solve the problem either. :-/

@vanzin vanzin deleted the conn-mgr-hang branch April 20, 2015 18:08
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants