[
https://issues.apache.org/jira/browse/TEZ-2917?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14983531#comment-14983531
]
Sergey Shelukhin commented on TEZ-2917:
---------------------------------------
MRInputLegacy deferring initialization - how is this useful. Another thing btw
is that whatever developer can figure out by just looking at config file,
metadata, etc., and code should not be on info level. I.e. logging configs and
other such things.
getContext().getDestinationVertexName() + ": "
+ "outputFormat=" + outputFormatClassName
+ ", using newmapreduce API=" + useNewApi - that looks like it could
just be figured out from the job and config
Using oldApi, MRpartitionerClass= - same
Waiting for N... - same. It is just logged all the time and it seems extremely
situational, when some small piece of code got stuck. Maybe it can log waiting
once.
Cleaning up task, Initializing task are extremely situational given the
surrounding log lines (i.e. running... is logged at info).
I dunno, I don't really care either way, this is based on users' complaints
that too much obscure stuff that doesn't matter is getting logged. [~hagleitn]
any opinion?
> change some logs from info to debug
> -----------------------------------
>
> Key: TEZ-2917
> URL: https://issues.apache.org/jira/browse/TEZ-2917
> Project: Apache Tez
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Attachments: TEZ-2917.patch
>
>
> I've done a highly unscientific summarization of the logs from some random
> queries, and will now change some log statements that are the most prevalent
> and not extremely useful from info to debug.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
