Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
MapReduce >> mail # user >> How to control of the output of "/stacks"


+
Shinichi Yamashita 2013-07-12, 21:54
Copy link to this message
-
Re: How to control of the output of "/stacks"
Thank you for comment.
I understand that stacks is very useful for debug and trouble-shooting.
I think that the change of LogLevel has a big influence on others.
So the control of the output to log file of stack traces is necessary.
I'll register patch with JIRA.

Regards,
Shinichi Yamashita

(2013/07/12 19:24), Harsh J wrote:
> The logging has sometimes come useful in debugging (i.e. if the stack
> on the UI went uncaptured, the log helps). It is currently not
> specifically toggle-able. I suppose it is OK to set it as DEBUG
> though. Can you file a JIRA for that please?
>
> The only way you can disable it right now is by (brute-forcibly)
> adding the below to the daemon's log4j.properties:
>
> log4j.logger.org.apache.hadoop.http.HttpServer=WARN
>
> Which may not be so ideal as we may miss other INFO from HttpServer generically.
>
> On Sat, Jul 13, 2013 at 3:24 AM, Shinichi Yamashita
> <[EMAIL PROTECTED]> wrote:
>> Hi,
>>
>> I can see the stack trace of the node when I access "/stacks" of Web UI.
>> And stack trace is output in the log file of the node, too.
>> Because the expansion of the log file and hard to see it, I don't want
>> to output it in a log file.
>> Is there the method to solve this problem?
>>
>> Regards,
>> Shinichi Yamashita
>
>
>