You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-issues@hadoop.apache.org by "Ayush Saxena (Jira)" <ji...@apache.org> on 2022/07/19 11:47:00 UTC

[jira] [Commented] (HDFS-14291) EC: GetServerDefaults To Support Default Erasure Coding Policy

    [ https://issues.apache.org/jira/browse/HDFS-14291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17568499#comment-17568499 ] 

Ayush Saxena commented on HDFS-14291:
-------------------------------------

Ahh, just bumped on this one and saw there is a question which I didn't answer since last 2.5 years. :) 

Well I now don't exactly remember all the use cases, but we then backported this to our internal fork.

Guess, one reason was when we set the default EC policy on a file, kind of send null for policyName in setErasureCoding policy, it then sets the default policy on the directory. 

So, some client application use case was there from some customer, so they wanted to know what is the default EC policy, rather than first set it and then do a get policy and if it is not the one they wanted and then change it. Well I can think of ways to do this without this today, but very faint memory why this.

 

Second was some upgrade scenario, when a new client tries to connect to an old server, which doesn't support ErasureCoding, it was something from the yarn side, that the application tries to unset any erasure coding policy set on the tmp or the staging directory. So, if the server doesn't support erasure coding policy(new client-old server setup), it was giving NoSuchMethod or so. So, the ask was to if the EC policy is null in getServerDefaults() that means server doesn't support EC policy, so don't shoot EC clean up RPCs and fail in the entire application flow, and assume there is no EC policy set anywhere...

> EC: GetServerDefaults To Support Default Erasure Coding Policy
> --------------------------------------------------------------
>
>                 Key: HDFS-14291
>                 URL: https://issues.apache.org/jira/browse/HDFS-14291
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>            Reporter: Ayush Saxena
>            Assignee: Ayush Saxena
>            Priority: Major
>         Attachments: HDFS-14291-01.patch
>
>




--
This message was sent by Atlassian Jira
(v8.20.10#820010)

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