[
https://issues.apache.org/jira/browse/HBASE-3283?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jonathan Gray updated HBASE-3283:
---------------------------------
Attachment: HBASE-3283-v1.patch
Adds null check and early out if so.
> NPE in AssignmentManager if processing shutdown of RS who doesn't have any
> regions assigned to it
> -------------------------------------------------------------------------------------------------
>
> Key: HBASE-3283
> URL: https://issues.apache.org/jira/browse/HBASE-3283
> Project: HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.90.0
> Reporter: Jonathan Gray
> Assignee: Jonathan Gray
> Fix For: 0.90.0, 0.92.0
>
> Attachments: HBASE-3283-v1.patch
>
>
> When doing server shutdown, we do the following:
> {noformat}
> deadRegions = new TreeSet<HRegionInfo>(this.servers.remove(hsi));
> {noformat}
> This removes the list of regions currently assigned to the specified server
> from the master in-memory state. But it's possible that the remove returns
> null (have seen it in running production cluster).
> We should check if this is null and just return if so.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
