Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Date: Wed, 16 May 2012 08:35:02 +0000 (UTC)
From: "chunhui shen (JIRA)" <jira@apache.org>
To: issues@hbase.apache.org
Message-ID: 
 <749522799.3357.1337157302880.JavaMail.tomcat@hel.zones.apache.org>
In-Reply-To: 
 <68467124.22364.1336060735146.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Commented] (HBASE-5927) SSH and DisableTableHandler
 happening together does not clear the znode of the region and RIT map.
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HBASE-5927?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13276592#comment-13276592 ] 

chunhui shen commented on HBASE-5927:
-------------------------------------

bq.But why should an assign happen and then go with reassign when we already know it is a disabling table

Yes, we needn't do the reassign. But I think it's the problem of ServerShutdownHandler#processDeadRegion,
why not return false for disabling table region.
So I think we could modify a little for the ServerShutdownHandler#processDeadRegion:
{code}
public static boolean processDeadRegion(HRegionInfo hri, Result result,
      AssignmentManager assignmentManager, CatalogTracker catalogTracker)
  throws IOException {
   ...
    if (hri.isOffline() && hri.isSplit()) {
      LOG.debug("Offlined and split region " + hri.getRegionNameAsString() +
        "; checking daughter presence");
      if (MetaReader.getRegion(catalogTracker, hri.getRegionName()) == null) {
        return false;
      }
      fixupDaughters(result, assignmentManager, catalogTracker);
      return false;
    }
    // If table is not disabled but the region is offlined,
    boolean disabling = assignmentManager.getZKTable().isDisablingTable(
        hri.getTableNameAsString());
    if (disabling) {
      LOG.info("The table " + hri.getTableNameAsString()
          + " is disabling.  Hence not assign it.");
      return false;
    }
    return true;
  }
{code}
                
> SSH and DisableTableHandler happening together does not clear the znode of the region and RIT map.
> --------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-5927
>                 URL: https://issues.apache.org/jira/browse/HBASE-5927
>             Project: HBase
>          Issue Type: Bug
>          Components: master
>    Affects Versions: 0.92.1, 0.96.0, 0.94.1
>            Reporter: Jieshan Bean
>            Assignee: Jieshan Bean
>             Fix For: 0.92.2, 0.96.0, 0.94.1
>
>         Attachments: HBASE-5927_94.patch, HBASE-5927_94_v2.patch, HBASE-5927_trunk.patch, HBASE-5927_trunk_2.patch, TestCaseForReProduce.txt
>
>
> A possible exception: If the related regionserver was just killed(But HMaster has not perceived that), then we will get a local exception "Connection reset by peer". If this region belongs to a disabling table. what will happen?
> ServerShutdownHandler will remove this region from AM#regions. So this region is still existing in RIT. TimeoutMonitor will take care of it after it got timeout. Then invoke unassign again. Since this region has been removed from AM#regions, it will return directly due to the below code:
> {code}
>     synchronized (this.regions) {
>       // Check if this region is currently assigned
>       if (!regions.containsKey(region)) {
>         LOG.debug("Attempted to unassign region " +
>           region.getRegionNameAsString() + " but it is not " +
>           "currently assigned anywhere");
>         return;
>       }
>     }
> {code}
> Then it leads to an end-less loop.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira