hadoop-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Harsh J <ha...@cloudera.com>
Subject Re: mapreduce.job.end-notification settings
Date Fri, 09 Nov 2012 20:24:14 GMT
Hey Steven,

Nice catch - this may be a bug we should fix.

There are 4 relevant properties in MR2:

Attempts: "mapreduce.job.end-notification.retry.attempts" and
"mapreduce.job.end-notification.max.attempts".

Retries: "mapreduce.job.end-notification.retry.interval" and
"mapreduce.job.end-notification.max.retry.interval".

I believe the configs of the latter of both of the above
classifications were meant to be added in as replacement names, but
the property names got added in wrong (as the former/older named ones)
in the XML.

Could you file a JIRA for this as well as there being an issue with
the word "seconds" in the description of retries? The code in MR2's
JobEndNotifier seems to expect seconds but uses it directly in
Thread.sleep(…) without making it milliseconds, which may be a bug we
need to fix as well, perhaps in a same issue as the configs ones.

On Fri, Nov 9, 2012 at 11:21 PM, Steven Willis <swillis@compete.com> wrote:
> I was looking at:
>
> http://hadoop.apache.org/docs/current/hadoop-mapreduce-client/hadoop-mapreduce-client-core/mapred-default.xml
>
> And I noticed that there are some duplicate properties with different values and different
descriptions:
>
> mapreduce.job.end-notification.retry.attempts   0       Indicates how many times hadoop
should attempt to contact the notification URL
> mapreduce.job.end-notification.retry.interval   30000   Indicates time in milliseconds
between notification URL retry calls
> ...
> mapreduce.job.end-notification.retry.attempts   5       The number of times the submitter
of the job wants to retry job end notification if it fails. This is capped by mapreduce.job.end-notification.max.attempts
> mapreduce.job.end-notification.retry.interval   1       The number of seconds the submitter
of the job wants to wait before job end notification is retried if it fails. This is capped
by mapreduce.job.end-notification.max.retry.interval
>
> Any idea which values and descriptions are correct?
>
> -Steve



-- 
Harsh J

Mime
View raw message