[
https://issues.apache.org/jira/browse/HADOOP-13159?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Xiaoyu Yao updated HADOOP-13159:
--------------------------------
Resolution: Fixed
Hadoop Flags: Reviewed
Fix Version/s: 2.8.0
Status: Resolved (was: Patch Available)
Thanks [~arpitagarwal] and [~templedf] for the review. I commit the patch v02
to trunk, branch-2 and branch-2.8 based on [~arpitagarwal]'s +1.
There is no unit test added for this change as it is difficult to test the race
between the main thread and the updater thread during initialization. I
manually validate the fix with multiple namenode restart and check the namenode
log, jmx and jstack to ensure there is no NPE and the jmx output as expected.
> Fix potential NPE in Metrics2 source for DecayRpcScheduler
> ----------------------------------------------------------
>
> Key: HADOOP-13159
> URL: https://issues.apache.org/jira/browse/HADOOP-13159
> Project: Hadoop Common
> Issue Type: Bug
> Components: ipc
> Affects Versions: 2.8.0
> Reporter: Xiaoyu Yao
> Assignee: Xiaoyu Yao
> Fix For: 2.8.0
>
> Attachments: HADOOP-13159.00.patch, HADOOP-13159.01.patch,
> HADOOP-13159.02.patch
>
>
> HADOOP-12985 introduced a few metrics2 counters for DecayRpcScheduler. The
> counters should be initialized before the updater thread is created/executed.
> Otherwise, this is a chance of NPE if the updater thread get executed before
> the counters are fully initialized. I will post a patch for it shortly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
