Subject: | |
From: | |
Reply To: | |
Date: | Tue, 20 Jun 2006 07:56:59 -0400 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
His processes on the compute nodes are allocated through PBS, so they should
be OK. The process on the head node is just telling them what to work on
next.
Dave
-----Original Message-----
From: Research Computing Support [mailto:[log in to unmask]] On
Behalf Of Robin
Sent: Monday, June 19, 2006 5:27 PM
To: [log in to unmask]
Subject: Re: process on head node
Dave,
Please kindly let the user knows that every hour we clean up compute
node processes. If he doesn't alllocate the nodes, his testing
processes will be killed.
So, if he's testing for a short time, it probably doesn't matter.
Thanks,
Robin
On Jun 19, 2006, at 3:53 PM, David Woods wrote:
> One user is running a slightly unusual process on the cluster. The
> user is
> Mike Helmick (helmicmt) - he is running what look like parallel
> jobs, but he
> also has a small Java process running on the head nodes. His
> processes on
> the compute nodes contact this process get their work units. He is
> doing
> this because his code is written in Java and couldn't get the mpiJava
> implementation to work on the cluster.
>
> I have told him about the 2 hour CPU time limit on the head node
> and his
> processes shouldn't have a problem with it.
>
> Dave
|
|
|