hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Enis Soztutar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-15093) Replication can report incorrect size of log queue for the global source when multiwal is enabled
Date Wed, 20 Jan 2016 03:05:40 GMT

    [ https://issues.apache.org/jira/browse/HBASE-15093?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15107888#comment-15107888
] 

Enis Soztutar commented on HBASE-15093:
---------------------------------------

bq. I saw this happening on a real production cluster. The size of global replication log
queue is almost never sum of all queues and some times goes negative.
Alright, was just curious. 

The failed tests seems relevant though. 

> Replication can report incorrect size of log queue for the global source when multiwal
is enabled
> -------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-15093
>                 URL: https://issues.apache.org/jira/browse/HBASE-15093
>             Project: HBase
>          Issue Type: Bug
>          Components: Replication
>    Affects Versions: 2.0.0, 1.2.0, 1.2.1
>            Reporter: Ashu Pachauri
>            Assignee: Ashu Pachauri
>            Priority: Minor
>         Attachments: HBASE-15093-V0.patch
>
>
> Replication can  report incorrect size for the size of log queue for the global source
when multiwal is enabled. This happens because the method MetricsSource#setSizeofLogQueue
performs non-trivial operations in a multithreaded world, even though it is not synchronized.

> We can simply divide MetricsSource#setSizeofLogQueue into MetricsSource#incrSizeofLogQueue
and MetricsSource#decrSizeofLogQueue. Not sure why we are currently directly setting the size
instead of incrementing/decrementing it.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message