[
https://issues.apache.org/jira/browse/HBASE-2513?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-2513:
-------------------------
Attachment: 2513-v3.patch
Thanks for the review Benoit.
This version fixes the indentation and adds some comments that hopefully will
alleviate suspicions. While true serveraddress is not needed inside the
method, it seemed sillier to me exposing the internals of how the processing of
events happens up in the parent method -- i.e. caller knowing about delay
queue -- hence passing of root region location and letting the process method
use it as it wants.
This patch is a fix for a bug introduced by 2414. 2414 started a refactoring
of master queue processing moving it out of master. Its not done. 2512 is
about finishing the work. There is a ways to go yet before queue processing is
standalone testable. This fix is about putting back a test that I left out
when I did 2414 (It was testing root region location down in the process of
operations method). As I see it, by the time 2512 is done, the queue
processing will likely make use of root location and more info from master
context that I'll need to pass in.
I'd suggest that this patch is good for the moment addressing the spin you were
seeing yesterday.
Thanks.
> hbase-2414 added bug where we'd tight-loop if no root available.
> ----------------------------------------------------------------
>
> Key: HBASE-2513
> URL: https://issues.apache.org/jira/browse/HBASE-2513
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: 2512-fix-for-2414-bug.patch, 2513-v2.patch, 2513-v3.patch
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
