You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "stack (JIRA)" <ji...@apache.org> on 2010/09/22 08:30:32 UTC
[jira] Created: (HBASE-3026) Fixup of "missing" daughters on split
is too aggressive
Fixup of "missing" daughters on split is too aggressive
-------------------------------------------------------
Key: HBASE-3026
URL: https://issues.apache.org/jira/browse/HBASE-3026
Project: HBase
Issue Type: Bug
Components: master
Reporter: stack
Fix For: 0.90.0
There is a bug in how we check for whether the daughters mentioned in parent region are present in .META. The check is done when we are processing a server shutdown. We're making the mistake of checking for presence of the daughter in the list of regions that used to live on the crashed server BUT fact of the matter is is that the daughter could just as well be rebalanced to another server.
The upshot is that we are inserting into .META. and trying to assign regions that are already assigned.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-3026) Fixup of "missing" daughters on split
is too aggressive
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-3026?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack updated HBASE-3026:
-------------------------
Attachment: 3026.txt
Here''s a patch that goes back to the META to check if daughters are present.
Testing it now.
> Fixup of "missing" daughters on split is too aggressive
> -------------------------------------------------------
>
> Key: HBASE-3026
> URL: https://issues.apache.org/jira/browse/HBASE-3026
> Project: HBase
> Issue Type: Bug
> Components: master
> Reporter: stack
> Fix For: 0.90.0
>
> Attachments: 3026.txt
>
>
> There is a bug in how we check for whether the daughters mentioned in parent region are present in .META. The check is done when we are processing a server shutdown. We're making the mistake of checking for presence of the daughter in the list of regions that used to live on the crashed server BUT fact of the matter is is that the daughter could just as well be rebalanced to another server.
> The upshot is that we are inserting into .META. and trying to assign regions that are already assigned.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (HBASE-3026) Fixup of "missing" daughters on split
is too aggressive
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-3026?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack resolved HBASE-3026.
--------------------------
Resolution: Fixed
Committed. Ran on cluster and cleaned it cleaned up the mess being made by incorrect daughter insertion/reassign.
> Fixup of "missing" daughters on split is too aggressive
> -------------------------------------------------------
>
> Key: HBASE-3026
> URL: https://issues.apache.org/jira/browse/HBASE-3026
> Project: HBase
> Issue Type: Bug
> Components: master
> Reporter: stack
> Fix For: 0.90.0
>
> Attachments: 3026.txt
>
>
> There is a bug in how we check for whether the daughters mentioned in parent region are present in .META. The check is done when we are processing a server shutdown. We're making the mistake of checking for presence of the daughter in the list of regions that used to live on the crashed server BUT fact of the matter is is that the daughter could just as well be rebalanced to another server.
> The upshot is that we are inserting into .META. and trying to assign regions that are already assigned.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.