hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kihwal Lee (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HDFS-4714) Log short messages in Namenode RPC server for exceptions meant for clients
Date Thu, 23 May 2013 20:39:23 GMT

     [ https://issues.apache.org/jira/browse/HDFS-4714?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Kihwal Lee updated HDFS-4714:

    Attachment: hdfs-4714.patch.txt

LeaseExpiredException is internal and few others are only thrown for admin commands.

I am pretty sure the overhead of string lookup in a hash map on call failures is cheaper than
logging full stack trace.
> Log short messages in Namenode RPC server for exceptions meant for clients
> --------------------------------------------------------------------------
>                 Key: HDFS-4714
>                 URL: https://issues.apache.org/jira/browse/HDFS-4714
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: namenode
>    Affects Versions: 3.0.0, 0.23.7, 2.0.5-beta
>            Reporter: Kihwal Lee
>         Attachments: hdfs-4714.branch-0.23.patch.txt, hdfs-4714.branch-2.patch.txt, hdfs-4714.patch.txt
> Namenode can slow down significantly if a rogue client/job issues massive number of requests
that will fail. E.g. permission denied, quota overage, etc.  The major contributing factor
in slow down is the long namenode log message, which includes full stack trace.  
> Previously similar issues involving safe mode and standby node have been addressed and
we can extend it to suppress logging stack traces for configured list of exception classes.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message