You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by "stack (JIRA)" <ji...@apache.org> on 2009/08/01 18:01:15 UTC
[jira] Created: (HBASE-1732) Flag to disable regionserver restart
Flag to disable regionserver restart
------------------------------------
Key: HBASE-1732
URL: https://issues.apache.org/jira/browse/HBASE-1732
Project: Hadoop HBase
Issue Type: Bug
Reporter: stack
Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1732) Flag to disable regionserver restart
Posted by "Andrew Purtell (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12737933#action_12737933 ]
Andrew Purtell commented on HBASE-1732:
---------------------------------------
Restart doesn't work for me either. RS log entries are misleading. They imply that the RS will restart but it does not. I imagine that will come up again and again on hbase-users@. I think we need a new RC which includes the patch for this issue with abort as default behavior until this is sorted for 0.20.1.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-1732) Flag to disable regionserver restart
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack updated HBASE-1732:
-------------------------
Attachment: norestart.patch
Here's a suggestion. Set the flag "hbase.regionserver.restart.on.zk.expire" in hbase-site.xml.
It doesn't abort, it just tries to close which probably isn't the right thing to do.... Abort seems extreme with its no flush of memstore content.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1732) Flag to disable regionserver restart
Posted by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12737908#action_12737908 ]
Jean-Daniel Cryans commented on HBASE-1732:
-------------------------------------------
Ok good, +1 on patch. Does it sinks RC1?
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-1732) Flag to disable regionserver restart
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack updated HBASE-1732:
-------------------------
Fix Version/s: 0.20.0
Bringing into 0.20.0.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Fix For: 0.20.0
>
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1732) Flag to disable regionserver restart
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12737910#action_12737910 ]
stack commented on HBASE-1732:
------------------------------
On sinking the RC, I'm wrong person to ask. I'm gung ho on getting 0.20.0 out there promptly unless a real showstopper.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (HBASE-1732) Flag to disable regionserver restart
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack resolved HBASE-1732.
--------------------------
Resolution: Fixed
Hadoop Flags: [Reviewed]
Committed. Thanks for review J-D.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Fix For: 0.20.0
>
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1732) Flag to disable regionserver restart
Posted by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12737905#action_12737905 ]
Jean-Daniel Cryans commented on HBASE-1732:
-------------------------------------------
The problem with flushing the memstore content is that the regions still contained in that RS may already be assigned to another one or, most likely, that the Master is splitting its logs. So I think it should release its resources as fast as possible.
Also I'm wondering, why disabling the restart code? The issue doesn't give the reason. Wouldn't it be better to remove it? Or do we keep it in order to make it better in the near future?
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Issue Comment Edited: (HBASE-1732) Flag to disable
regionserver restart
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12737907#action_12737907 ]
stack edited comment on HBASE-1732 at 8/1/09 9:16 AM:
------------------------------------------------------
Sorry, I should have said.... jgray asked for this mechanism ("Anything for..")... He was saying restart is just broken for him out on streamy. Need to investigate more.
Yes on regions could already be deployed on another server... so v2 does abort instead.
was (Author: stack):
Sorry, I should have said.... jgray asked for this mechanism.
Yes on regions could already be deployed on another server... so v2 does abort instead.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-1732) Flag to disable regionserver restart
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack updated HBASE-1732:
-------------------------
Attachment: norestart-v2.patch
Sorry, I should have said.... jgray asked for this mechanism.
Yes on regions could already be deployed on another server... so v2 does abort instead.
> Flag to disable regionserver restart
> ------------------------------------
>
> Key: HBASE-1732
> URL: https://issues.apache.org/jira/browse/HBASE-1732
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Attachments: norestart-v2.patch, norestart.patch
>
>
> Add a flag so on zk session expiration, regionserver goes down and doesn't try restarting.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.