hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yu Li (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-14075) HBaseClusterManager should use port(if given) to find pid
Date Tue, 14 Jul 2015 18:04:04 GMT

    [ https://issues.apache.org/jira/browse/HBASE-14075?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14626781#comment-14626781

Yu Li commented on HBASE-14075:

After a second thought and some test, the changes in v1 patch on HBaseClusterManager#isRunningCommand
is incorrect, because in the real world it's possible that the rpc server is down(so no more
listening on rpc port) but the Master/RS is still in progress of shutting down.

So for methods which need to check whether the service is running, we should firstly get pid
of the service through port and name, then use the pid to check whether service is still running.
Please refer to the patch for the detailed changes

> HBaseClusterManager should use port(if given) to find pid
> ---------------------------------------------------------
>                 Key: HBASE-14075
>                 URL: https://issues.apache.org/jira/browse/HBASE-14075
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Yu Li
>            Assignee: Yu Li
>            Priority: Minor
>         Attachments: HBASE-14075.patch
> This issue is found while we run ITBLL in distributed cluster. Our testing env is kind
of special that we run multiple regionserver instance on a single physical machine, so {noformat}ps
-ef | grep proc_regionserver{noformat} will return more than one line, thus cause the tool
might check/kill the wrong process
> Actually in HBaseClusterManager we already introduce port as a parameter for methods
like isRunning, kill, etc. So the only thing to do here is to get pid through port if port
is given

This message was sent by Atlassian JIRA

View raw message