hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Rakesh R (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-11428) Change setErasureCodingPolicy to take a required string EC policy name
Date Mon, 27 Feb 2017 10:28:45 GMT

    [ https://issues.apache.org/jira/browse/HDFS-11428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15885522#comment-15885522
] 

Rakesh R commented on HDFS-11428:
---------------------------------

Thanks [~andrew.wang] for the useful improvement. Overall patch looks good to me. I've few
suggestions,
# Its good to update javadoc with {{@throws IllegalArgumentException if the given ecPolicyName
is invalid}}.
# Minor suggestion to validate null {{ecPolicyName}}, otw it will throw NPE.
{code}
DistributedFileSystem#setErasureCodingPolicy()

    if (ecPolicyName == null) {
      throw new IOException("Invalid erasure coding policy name");
    }
{code}

> Change setErasureCodingPolicy to take a required string EC policy name
> ----------------------------------------------------------------------
>
>                 Key: HDFS-11428
>                 URL: https://issues.apache.org/jira/browse/HDFS-11428
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: erasure-coding
>    Affects Versions: 3.0.0-alpha1
>            Reporter: Andrew Wang
>            Assignee: Andrew Wang
>              Labels: hdfs-ec-3.0-must-do
>         Attachments: HDFS-11428.001.patch, HDFS-11428.002.patch, HDFS-11428.003.patch
>
>
> The current {{setErasureCodingPolicy}} API takes an optional {{ECPolicy}}. This makes
calling the API harder for clients, since they need to turn a specified name into a policy,
and the set of available EC policies is only available on the NN.
> You can see this awkwardness in the current EC cli set command: it first fetches the
list of EC policies, looks for the one specified by the user, then calls set. This means we
need to issue two RPCs for every set (inefficient), and we need to do validation on the NN
side anyway (extraneous work).
> Since we're phasing out the system default EC policy, it also makes sense to make the
policy a required parameter.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org


Mime
View raw message