You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Gary Helmling (JIRA)" <ji...@apache.org> on 2011/09/23 21:45:26 UTC
[jira] [Created] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
MiniHBaseCluster.shutdown() doesn't work if no active master
------------------------------------------------------------
Key: HBASE-4472
URL: https://issues.apache.org/jira/browse/HBASE-4472
Project: HBase
Issue Type: Bug
Reporter: Gary Helmling
Attachments: HBASE-4472.patch
Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Assigned] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Ted Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ted Yu reassigned HBASE-4472:
-----------------------------
Assignee: Gary Helmling
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Assignee: Gary Helmling
> Fix For: 0.92.0
>
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Eugene Koontz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13113784#comment-13113784 ]
Eugene Koontz commented on HBASE-4472:
--------------------------------------
I think Gary's patch is better. It's simpler - with mine, you have the master modifying things outside itself (i.e. a zookeeper node) before aborting.
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Eugene Koontz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Eugene Koontz updated HBASE-4472:
---------------------------------
Attachment: jstack.txt
jstack of test during hang: note the TIMED_WAITING in HRegionServer.getMaster().
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Gary Helmling (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Gary Helmling updated HBASE-4472:
---------------------------------
Attachment: HBASE-4472.patch
Adds an explicit call to HRegionServer.stop() in JVMClusterUtil.shutdown().
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Attachments: HBASE-4472.patch
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Ted Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13113806#comment-13113806 ]
Ted Yu commented on HBASE-4472:
-------------------------------
I applied Gary's patch to 0.92 and TRUNK.
Thanks Gary and Eugene.
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Eugene Koontz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Eugene Koontz updated HBASE-4472:
---------------------------------
Attachment: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch
Call this.clusterStatusTracker.setClusterDown() during HMaster.abort().
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Eugene Koontz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13113724#comment-13113724 ]
Eugene Koontz commented on HBASE-4472:
--------------------------------------
HBASE-4472.delete-zk-cluster-node-on-master-abort.patch is an alternative patch to fix the same bug: Master will remove the cluster-status ZK node (/hbase/shutdown by default) when it aborts. This will prevent the HRegionServer from hanging in HRegionServer.getMaster().
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Ted Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ted Yu resolved HBASE-4472.
---------------------------
Resolution: Fixed
Hadoop Flags: [Reviewed]
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Assignee: Gary Helmling
> Fix For: 0.92.0
>
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Ted Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ted Yu updated HBASE-4472:
--------------------------
Fix Version/s: 0.92.0
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Assignee: Gary Helmling
> Fix For: 0.92.0
>
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Hudson (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13113878#comment-13113878 ]
Hudson commented on HBASE-4472:
-------------------------------
Integrated in HBase-0.92 #18 (See [https://builds.apache.org/job/HBase-0.92/18/])
HBASE-4472 MiniHBaseCluster.shutdown() doesn't work if no active master
tedyu :
Files :
* /hbase/branches/0.92/CHANGES.txt
* /hbase/branches/0.92/src/main/java/org/apache/hadoop/hbase/util/JVMClusterUtil.java
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Assignee: Gary Helmling
> Fix For: 0.92.0
>
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4472) MiniHBaseCluster.shutdown() doesn't
work if no active master
Posted by "Hudson (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13113904#comment-13113904 ]
Hudson commented on HBASE-4472:
-------------------------------
Integrated in HBase-TRUNK #2247 (See [https://builds.apache.org/job/HBase-TRUNK/2247/])
HBASE-4472 MiniHBaseCluster.shutdown() doesn't work if no active master
tedyu :
Files :
* /hbase/trunk/CHANGES.txt
* /hbase/trunk/src/main/java/org/apache/hadoop/hbase/util/JVMClusterUtil.java
> MiniHBaseCluster.shutdown() doesn't work if no active master
> ------------------------------------------------------------
>
> Key: HBASE-4472
> URL: https://issues.apache.org/jira/browse/HBASE-4472
> Project: HBase
> Issue Type: Bug
> Reporter: Gary Helmling
> Assignee: Gary Helmling
> Fix For: 0.92.0
>
> Attachments: HBASE-4472.delete-zk-cluster-node-on-master-abort.patch, HBASE-4472.patch, jstack.txt
>
>
> Running tests over in HBASE-4014 brought up this issue. If the active master in a MiniHBaseCluster has aborted, then calling MiniHBaseCluster.shutdown() will just hang in JVMClusterUtil.shutdown(), waiting to join each of the region server threads.
> Seems like we should explicitly stop each region server instead of just relying on an active master instance deleting the cluster status znode?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira