hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jerome Boulon (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-5118) ChukwaAgent controller should retry to register for a longer period but not as frequent as now
Date Tue, 10 Feb 2009 00:48:59 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-5118?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12672090#action_12672090
] 

Jerome Boulon commented on HADOOP-5118:
---------------------------------------

Sure:-)
Watchdog is watching for ChukwaAgent only once every 5 minutes, so there's no point in retrying
more than once every 5 mins.

In practice, if the watchdog is not able to automatically restart the agent, it will take
more than 20 minutes to get Ops to restart it.
Also Ops want us to limit the number of communications between Hadoop and Chukwa, that's why
30 minutes.


> ChukwaAgent controller should retry to register for a longer period but not as frequent
as now 
> -----------------------------------------------------------------------------------------------
>
>                 Key: HADOOP-5118
>                 URL: https://issues.apache.org/jira/browse/HADOOP-5118
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: contrib/chukwa
>            Reporter: Jerome Boulon
>            Assignee: Jerome Boulon
>         Attachments: HADOOP-5118.patch, HADOOP-5118.patch
>
>
> if the agent is down, most chances are that either it will be up again not before 1 minute
(watchdog) or it will take longer
> So it's better to retry in 1 minute for the first time then try every 30 minutes for
the next 24 hours

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message