[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] [newb] setting up 2nd machine
- Date: Thu, 28 Jan 2010 13:41:25 -0500
- From: Matthew Farrellee <matt@xxxxxxxxxx>
- Subject: Re: [Condor-users] [newb] setting up 2nd machine
Neal Becker wrote:
Alan De Smet wrote:
Thanks. I installed from fedora rpm, which does not include
condor_configure. Should I try to get hold of a condor_configure, or can I
just manually set this up?
Neal Becker <ndbecker2@xxxxxxxxx> wrote:
I've got my first 8-core machine running fine. Setup on fedora 12 from
Now I'd like to add a second machine that acts as a compute server. Real
basic requirements - no parallel stuff.
I've tried reading the docs, but I'm still clueless. What do I need to
Based on your description, I'm assuming that your first machine
is running a schedd, a startd, a collector, and a negotiator.
That is, it's an entirely self-contained pool. Given that, your
next steps are:
0. Ensure that the two machines have unrestricted network access
to each other. (Condor can cope with firewalls and NATs, but
that's a more complex question.)
1. Install Condor on the second machine.
2. Configure Condor on the second machine to only run the startd.
(And the master. You always run a master.)
3. Configure Condor on the second machine to report to the
collector on the first machine.
4. Ensure that the first machine is configured to accept writes
from the second machine
5. Ensure that the second machine is configured to accept writes
from the first.
condor_configure/condor_install can probably handle steps 1-3 for
you. You'll want something like:
condor_configure --type=execute --central-manager=first.example.com
--type=execute will set it up to run the startd.
--central-manager=first-machine.com will have it report to the
collector on the first machine. Obviously change
"first.example.com" to the actual name of your first machine.
You can manually set it up, see http://kbase.redhat.com/faq/docs/DOC-23598