You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-user@hadoop.apache.org by Hamid Oliaei <ol...@gmail.com> on 2012/06/26 10:59:55 UTC

Understanding job completion in other nodes

Hi,

I want to run a job on all of nodes and if one job was completed, the node
must wait until the jobs on the other nodes finish.
For that, every node must signal to all nodes and when every node receives
the signal from every one, next job must be run.
How can I handle that in Hadoop?
Is there any solution to understanding job completion in other nodes?
P.S: I am using Hadoop 0.20.2.

Thanks,

Hamid Oliaei

Oliaei@gmail.com

Re: Understanding job completion in other nodes

Posted by Hamid Oliaei <ol...@gmail.com>.
Hi  Christoph ,

I didn't consider waitForCompletion. I'll try using that and hope my
workflow didn't need any additional method.

Thanks a lot.

Hamid Oliaei

Oliaei@gmail.com

AW: Understanding job completion in other nodes

Posted by Christoph Schmitz <ch...@1und1.de>.
Hi Hamid,

I'm not sure if I understand your question correctly, but I think this is exactly what the standard workflow in a Hadoop application looks like:

Job job1 = new Job(...);
// setup job, set Mapper and Reducer, etc. 
job1.waitForCompletion(...); // at this point, the cluster will run job 1 and wait for its completion

// think about the results from job 1, plan job 2 accordingly

Job job2 = new Job(...);
// setup another job
job2.waitForCompletion(...); // at this point, the cluster will run job 2 and wait for its completion

etc.

So I think "waitForCompletion" is what you were asking about, right?

Regards,
Christoph

-----Ursprüngliche Nachricht-----
Von: Hamid Oliaei [mailto:oliaei@gmail.com] 
Gesendet: Dienstag, 26. Juni 2012 11:00
An: mapreduce-user@hadoop.apache.org
Betreff: Understanding job completion in other nodes

Hi,

I want to run a job on all of nodes and if one job was completed, the node must wait until the jobs on the other nodes finish. 
For that, every node must signal to all nodes and when every node receives the signal from every one, next job must be run.
How can I handle that in Hadoop?
Is there any solution to understanding job completion in other nodes?
P.S: I am using Hadoop 0.20.2.

Thanks,     


Hamid Oliaei

Oliaei@gmail.com