hadoop-mapreduce-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Philip Zeyliger <phi...@cloudera.com>
Subject Re: Add user jars to mapreduce
Date Mon, 04 Jan 2010 18:20:33 GMT
Visit the machine that's running the job.  Every job has a working
directory, where the cache files should have been "localized".  See if the
file is there.

You might also look at the logs of the TaskTracker, to see if it's throwing
any errors during the localization process.

-- Philip

On Mon, Jan 4, 2010 at 10:11 AM, Eric Yang <eyang@yahoo-inc.com> wrote:

>
>
> >> Your above code should work properly. Added jars will not be part of
> job.jar.
> >> The addon.jar will be uploaded by the client, the added libjars will be
> added
> >> to classpath using DistributedCache.
> >> So, above code should work unless your path to jar or the jar itself was
> >> wrong.
>
> When the path is wrong or jar is corrupted, there are exceptions thrown.  I
> am not getting neither exception messages.  Therefore, I assume the jar
> file
> is loaded as expected.
>
> I also tried another method by using DistributedCache syntax to add jar
> from
> HDFS to jobconf, and run the job.  It didsn't work neither.
>
> Is there any procedure that I could follow to debug this further?
>
> Regards,
> Eric
>
>

Mime
View raw message