[ 
https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14518576#comment-14518576
 ] 

Vinod Kumar Vavilapalli commented on YARN-2942:
-----------------------------------------------

Tx for the updated docs, [~rkanter]!

The proposal really is a poor man's replacement for the absence of concurrency 
control in HDFS. The good thing about the proposal is that it is not shipping 
logs across the wire multiple times. The challenge is going to be fault 
handling. We need to make sure that there is someone centrally listening to 
node membership changes too (for e.g. to handle lost nodes).

It's sort of spelled out in the doc, but repeating for clarity: I am assuming 
that we still continue to write the per-node file and have an aggregated-file 
by the side. IAC, we should have a way for folks to alternate to this, with 
existing implementation as a backup.

Regarding log-aggregation status, YARN-1376 and friends added some support (I 
am reviewing them after the fact).

I am still interesting in pursuing variable-length files as an orthogonal 
feature. 

/cc [~jlowe], [~knoguchi] who have experience with log aggregation at large 
scale.

> Aggregated Log Files should be combined
> ---------------------------------------
>
>                 Key: YARN-2942
>                 URL: https://issues.apache.org/jira/browse/YARN-2942
>             Project: Hadoop YARN
>          Issue Type: New Feature
>    Affects Versions: 2.6.0
>            Reporter: Robert Kanter
>            Assignee: Robert Kanter
>         Attachments: CombinedAggregatedLogsProposal_v3.pdf, 
> CombinedAggregatedLogsProposal_v6.pdf, 
> CompactedAggregatedLogsProposal_v1.pdf, 
> CompactedAggregatedLogsProposal_v2.pdf, 
> ConcatableAggregatedLogsProposal_v4.pdf, 
> ConcatableAggregatedLogsProposal_v5.pdf, YARN-2942-preliminary.001.patch, 
> YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, 
> YARN-2942.003.patch
>
>
> Turning on log aggregation allows users to easily store container logs in 
> HDFS and subsequently view them in the YARN web UIs from a central place.  
> Currently, there is a separate log file for each Node Manager.  This can be a 
> problem for HDFS if you have a cluster with many nodes as you’ll slowly start 
> accumulating many (possibly small) files per YARN application.  The current 
> “solution” for this problem is to configure YARN (actually the JHS) to 
> automatically delete these files after some amount of time.  
> We should improve this by compacting the per-node aggregated log files into 
> one log file per application.



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

Reply via email to