hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Todd Lipcon (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-4275) RS should communicate fatal "aborts" back to the master
Date Tue, 30 Aug 2011 02:46:38 GMT

     [ https://issues.apache.org/jira/browse/HBASE-4275?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Todd Lipcon updated HBASE-4275:
-------------------------------

    Attachment: 0005-HBASE-4275.-Report-fatals-to-HMaster.patch

Here's a patch which adds the new RPC.

The master logs it locally and also keeps a ring-buffer of such log messages, constrained
to 1MB by default.

I've tested this on a cluster in coordination with HBASE-4292 (which displays the ring buffer)
and it's very useful for debugging.

The RS-abort behavior should probably get its own unit test before commit (or add some assertions
to existing tests)

> RS should communicate fatal "aborts" back to the master
> -------------------------------------------------------
>
>                 Key: HBASE-4275
>                 URL: https://issues.apache.org/jira/browse/HBASE-4275
>             Project: HBase
>          Issue Type: Improvement
>          Components: master, regionserver
>    Affects Versions: 0.92.0
>            Reporter: Todd Lipcon
>         Attachments: 0005-HBASE-4275.-Report-fatals-to-HMaster.patch
>
>
> When a region server aborts, it should attempt to send an RPC to the master that contains
(a) the reason for aborting, and (b) the last several KB of log messages, if available. This
should help a lot in debugging.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message