Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [HTCondor-users] MPI Dedicated Scheduler
- Date: Thu, 23 Apr 2015 11:19:06 -0500
- From: Greg Thain <gthain@xxxxxxxxxxx>
- Subject: Re: [HTCondor-users] MPI Dedicated Scheduler
On 04/23/2015 09:34 AM, Matt Mulqueen wrote:
Good morning,
I'm going through the steps to allow my users to submit MPI jobs and
need clarification regarding the dedicated scheduler. I believe that
this needs to be a different machine than the head node. Is this
correct? Currently, users submit all jobs from that node and it runs
the COLLECTOR, MASTER, NEGOTIATOR, and SCHEDD daemons. I need a
different machine to function as the dedicated scheduler, right? I
just want to be sure before going shopping for a new machine.
The dedicated scheduler code is part of the usual HTCondor scheduler.
While you can put this on a separate machine, perhaps to load balance,
if your current submit machine is not overloaded, you can submit and
manage parallel jobs from your existing scheduler machine. If you have
another machine available, we would first recommend to run the central
manager components (the negotiator and collector) on a separate machine
from the schedd, though that is not strictly required.
-greg