[jira] [Updated] (HDFS-6840) Clients are always sent to the same datanode when read is off rack

["Andrew Wang (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HDFS-6840?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Andrew Wang updated HDFS-6840:
------------------------------

    Attachment: hdfs-6840.001.patch

Here's a patch that removes the "random seed" stuff. Fortunately HDFS-6701 hasn't been released yet, so we don't need to do any config key deprecation.

branch-2 is probably going to need a new patch like the others. This might be a good time to try and figure out why we have those extra test files.

> Clients are always sent to the same datanode when read is off rack
> ------------------------------------------------------------------
>
>                 Key: HDFS-6840
>                 URL: https://issues.apache.org/jira/browse/HDFS-6840
>             Project: Hadoop HDFS
>          Issue Type: Bug
>    Affects Versions: 2.5.0
>            Reporter: Jason Lowe
>            Assignee: Andrew Wang
>            Priority: Critical
>         Attachments: hdfs-6840.001.patch
>
>
> After HDFS-6268 the sorting order of block locations is deterministic for a given block and locality level (e.g.: local, rack. off-rack), so off-rack clients all see the same datanode for the same block.  This leads to very poor behavior in distributed cache localization and other scenarios where many clients all want the same block data at approximately the same time.  The one datanode is crushed by the load while the other replicas only handle local and rack-local requests.



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