[
https://issues.apache.org/jira/browse/GIRAPH-1187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17177875#comment-17177875
]
Randy Hu commented on GIRAPH-1187:
----------------------------------
[~dlogothetis], thanks for the update. We happened to run large scale tests
recently and hit by the issue again.
My local fix is simply report only when gcTimeMetric is initialized by checking
whether it is null. It's not ideal, but sufficient from Giraph perspective
since it really cares metrics when the computation starts. The issue could only
happen in super step -1, which loads vertices/edges from storage. I can submit
a PR if that's acceptable.
> NPE in worker GC listener
> --------------------------
>
> Key: GIRAPH-1187
> URL: https://issues.apache.org/jira/browse/GIRAPH-1187
> Project: Giraph
> Issue Type: Bug
> Components: graph
> Affects Versions: 1.2.0
> Environment: Giraph 1.2.0 running in M/R on Hadoop 2.6
> Reporter: Randy Hu
> Priority: Critical
>
> Likely the GC notification happened before any super step, in which case "
> gcTimeMetric" is not initialized at all. Here is stack trace:
>
> java.lang.NullPointerException at
> org.apache.giraph.graph.GraphTaskManager$1.handleNotification(GraphTaskManager.java:667)
> at
> sun.management.NotificationEmitterSupport.sendNotification(NotificationEmitterSupport.java:156)
> at
> sun.management.GarbageCollectorImpl.createGCNotification(GarbageCollectorImpl.java:143)
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
