You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@storm.apache.org by agresch <gi...@git.apache.org> on 2018/08/29 17:43:49 UTC
[GitHub] storm pull request #2816: STORM-3208 fix worker kill NPE
GitHub user agresch opened a pull request:
https://github.com/apache/storm/pull/2816
STORM-3208 fix worker kill NPE
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/agresch/storm agresch_storm-3208
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/storm/pull/2816.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #2816
----
commit 91458a32cded87fb64dbb1b22db8059edd6af99d
Author: Aaron Gresch <ag...@...>
Date: 2018-08-29T17:42:55Z
STORM-3208 fix worker kill NPE
----
---
[GitHub] storm pull request #2816: STORM-3208 fix worker kill NPE
Posted by asfgit <gi...@git.apache.org>.
Github user asfgit closed the pull request at:
https://github.com/apache/storm/pull/2816
---
[GitHub] storm issue #2816: STORM-3208 fix worker kill NPE
Posted by revans2 <gi...@git.apache.org>.
Github user revans2 commented on the issue:
https://github.com/apache/storm/pull/2816
@agresch could we at least put in some logging when it is null?
---
[GitHub] storm issue #2816: STORM-3208 fix worker kill NPE
Posted by agresch <gi...@git.apache.org>.
Github user agresch commented on the issue:
https://github.com/apache/storm/pull/2816
@revans2 - this is a normal run of our integration test pipeline. I don't have much context, just noticed this error in the logs and filed the JIRA.
---
[GitHub] storm issue #2816: STORM-3208 fix worker kill NPE
Posted by agresch <gi...@git.apache.org>.
Github user agresch commented on the issue:
https://github.com/apache/storm/pull/2816
@danny0405 - I'm not really knowledgable about any of this particular code, but the LocalState API clearly allows returning null, rather than throwing any exception. I'm just trying to cover the callstack I provided in the JIRA. If we feel this shouldn't happen, maybe LocalState getApprovedWorkers() should throw an exception if there are no approved workers?
---
[GitHub] storm issue #2816: STORM-3208 fix worker kill NPE
Posted by danny0405 <gi...@git.apache.org>.
Github user danny0405 commented on the issue:
https://github.com/apache/storm/pull/2816
@agresch
So in what case `_localState.getApprovedWorkers()` will return null.
---
[GitHub] storm issue #2816: STORM-3208 fix worker kill NPE
Posted by revans2 <gi...@git.apache.org>.
Github user revans2 commented on the issue:
https://github.com/apache/storm/pull/2816
@danny0405 @agresch this should not happen under normal conditions. If it is happening under normal conditions we need to fix some things. The approved workers holds the list of workers that the supervisor believes it is currently running. The only times that it would become null would be when there is an issue with the local state, like it was wiped clean too quickly by an over aggressive script meant to fix a broken node, or by a test framework that possibly got the timing wrong.
If this is happening regularly outside of odd situations like I said, then there may be a race condition that we missed and we need to look into it more deeply.
---
[GitHub] storm issue #2816: STORM-3208 fix worker kill NPE
Posted by agresch <gi...@git.apache.org>.
Github user agresch commented on the issue:
https://github.com/apache/storm/pull/2816
@revans2 - added a warning.
---