hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vinayakumar B (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-6247) Avoid timeouts for replaceBlock() call by sending intermediate responses to Balancer
Date Wed, 23 Jul 2014 09:35:38 GMT

    [ https://issues.apache.org/jira/browse/HDFS-6247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14071555#comment-14071555
] 

Vinayakumar B commented on HDFS-6247:
-------------------------------------

Failure, even though related to Balancing, Its not caused by this patch.

In fact, its failed due to selection of a block belongs to "/system/balancer.id" for the movement
which is having default replication(3) and after movement it will not be detected as excess.
All other blocks in test having 1 replication.
So the calculation in TestBalancer#waitForBalancer(..) does not meet and test timesout. I
think this can be fixed in a separate jira if observed again.

Anyway, triggering the QA again.

> Avoid timeouts for replaceBlock() call by sending intermediate responses to Balancer
> ------------------------------------------------------------------------------------
>
>                 Key: HDFS-6247
>                 URL: https://issues.apache.org/jira/browse/HDFS-6247
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: balancer, datanode
>    Affects Versions: 2.4.0
>            Reporter: Vinayakumar B
>            Assignee: Vinayakumar B
>         Attachments: HDFS-6247.patch, HDFS-6247.patch, HDFS-6247.patch, HDFS-6247.patch
>
>
> Currently there is no response sent from target Datanode to Balancer for the replaceBlock()
calls.
> Since the Block movement for balancing is throttled, complete block movement will take
time and this could result in timeout at Balancer, which will be trying to read the status
message.
>  
> To Avoid this during replaceBlock() call in in progress Datanode  can send IN_PROGRESS
status messages to Balancer to avoid timeouts and treat BlockMovement as  failed.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message