You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@tez.apache.org by "Siddharth Seth (JIRA)" <ji...@apache.org> on 2014/10/21 01:39:34 UTC

[jira] [Commented] (TEZ-1424) Fixes to DAG text representation in debug mode

    [ https://issues.apache.org/jira/browse/TEZ-1424?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14177660#comment-14177660 ] 

Siddharth Seth commented on TEZ-1424:
-------------------------------------

Comments on the patch.
- Rather than printing specific fields, I think it's better to print all fields except credentials. Can be done by walking the field descriptors. Walking the descriptors should allow this. That also ends up printing the name of the field correctly. e.g. Sleep jobname: "initialmap" - that's mixing the job name and the vertex name.
- For the credentials itself, we should parse them out - print counts, as well as the Token Identifier (which gives type information - HDFS / session / etc).
- {code}
-    submittedDAGs.incrementAndGet();{code}
Accidental removal ? Also, the ordering is changed where we log the debug after starting the DAG.

> Fixes to DAG text representation in debug mode
> ----------------------------------------------
>
>                 Key: TEZ-1424
>                 URL: https://issues.apache.org/jira/browse/TEZ-1424
>             Project: Apache Tez
>          Issue Type: Bug
>            Reporter: Siddharth Seth
>            Priority: Critical
>         Attachments: TEZ-1424.1.patch
>
>
> Several fixes required
> - Don't log entire tokens, just the identifier should be enough
> - DAG ID (or unique number needs to be used). Otherwise we get only one file per session
> - This should not go into the local-directory - since that isn't accessible. Instead the log directory would be a better place.
> Marking as critical for 0.5.1 since this is very useful for debugging.



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