hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Wang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-3702) Add an option for NOT writing the blocks locally if there is a datanode on the same box as the client
Date Mon, 21 Mar 2016 20:55:25 GMT

    [ https://issues.apache.org/jira/browse/HDFS-3702?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15205111#comment-15205111
] 

Andrew Wang commented on HDFS-3702:
-----------------------------------

[~stack] from a downstream perspective, could you comment on the usability of providing node
lists to this API? This always felt hacky to me, since ultimately the NN is the one who knows
about the cluster state and DN names and block location constraints. My impression was that
tracking this in HBase was onerous, and is part of why favored nodes fell out of favor.

bq. For example, some application may want to distribute its files uniformly in a cluster....

The main reason for skew I've seen is the local writer case, which this patch attempts to
address. It'll still bias to the local rack, but I doubt that'll be an issue in practice,
and if it is we can also add another flag for fully random distribution.

> Add an option for NOT writing the blocks locally if there is a datanode on the same box
as the client
> -----------------------------------------------------------------------------------------------------
>
>                 Key: HDFS-3702
>                 URL: https://issues.apache.org/jira/browse/HDFS-3702
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: hdfs-client
>    Affects Versions: 2.5.1
>            Reporter: Nicolas Liochon
>            Assignee: Lei (Eddy) Xu
>            Priority: Minor
>              Labels: BB2015-05-TBR
>         Attachments: HDFS-3702.000.patch, HDFS-3702.001.patch, HDFS-3702.002.patch, HDFS-3702.003.patch,
HDFS-3702.004.patch, HDFS-3702.005.patch, HDFS-3702.006.patch, HDFS-3702.007.patch, HDFS-3702.008.patch,
HDFS-3702_Design.pdf
>
>
> This is useful for Write-Ahead-Logs: these files are writen for recovery only, and are
not read when there are no failures.
> Taking HBase as an example, these files will be read only if the process that wrote them
(the 'HBase regionserver') dies. This will likely come from a hardware failure, hence the
corresponding datanode will be dead as well. So we're writing 3 replicas, but in reality only
2 of them are really useful.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message