hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Matt Foley (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-2114) re-commission of a decommissioned node does not delete excess replica
Date Tue, 19 Jul 2011 18:00:59 GMT

    [ https://issues.apache.org/jira/browse/HDFS-2114?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13067866#comment-13067866
] 

Matt Foley commented on HDFS-2114:
----------------------------------


Hi John, big improvement.  Comments:

1. Thanks for finding the place where decommissioned nodes are sorted to the end of the list.
 I feel much better now :-)

2. I'll go along with use of the isNodeDown flag on the grounds that if a node name were ever
null (which should be impossible) it might match a null value of "downnode".

3. This is a nit, but could you please change local constant NAMENODE_REPLICATION_INTERVAL_KEY
to NAMENODE_REPLICATION_INTERVAL ?

4. checkFile():  You've definitely improved the inner loop a lot.  However, on second reading
I am concerned about the use of checkFile() in testRecommission(). I believe the use in testDecommission()
is intended to be instantaneous, reading the state of the system before replica deletion has
time to be done, while the use in testRecommission() is intended to be after it reaches steady
state, after the system has readjusted the number of replicas.  The use of exceptions to notify
problem states doesn't work too well in the latter case, as you had to work around.  Also
I'm concerned about race conditions while trying to get a particular reading from checkFile()
during potential changes in replication state.  My suggested fix is too long for a comment,
so I've emailed you a code fragment.  Thanks.


> re-commission of a decommissioned node does not delete excess replica
> ---------------------------------------------------------------------
>
>                 Key: HDFS-2114
>                 URL: https://issues.apache.org/jira/browse/HDFS-2114
>             Project: Hadoop HDFS
>          Issue Type: Bug
>            Reporter: John George
>            Assignee: John George
>         Attachments: HDFS-2114-2.patch, HDFS-2114-3.patch, HDFS-2114-4.patch, HDFS-2114-5.patch,
HDFS-2114.patch
>
>
> If a decommissioned node is removed from the decommissioned list, namenode does not delete
the excess replicas it created while the node was decommissioned.

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

        

Mime
View raw message