[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Question about using Condor in anon-Grid environment

From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Greg Thain
> This will create one parallel job with two "procs", the first of which 
> requires 3 windows machines, the other, 7 linux machines.  Condor will 
> wait until all 10 machines are available, then launch the job on those 
> machines.  (Of course, the jobs will not launch on each machine 
> completely simultaneously, so you may need to be aware of race 
> conditions at startup).

Thanks, that helps. However, in our case we don't want Condor to start anything on the servers and clients, we just want it to notify our "in-house tool" that the resources are now reserved.

That's the part where I'm not sure if Condor is the right solution. I understand that it provides a lot of cluster-related functionalities and can manage dependencies between jobs, but we don't need most of that. We just want to use the resource management / job queue part of it. That's also why I mentioned that we ideally don't want to install anything on the servers / clients. It should just take the available nodes out of a database, then mark them reserved or unreserved and then hand over these resources to us. Example:

- User1 submits a job to Condor requesting 3 linux clients, 2 windows clients and 2 servers.
- Condor queues the job until the resources become available (assuming there are other Condor jobs running and insufficient resources are currently available).
- Once the resources are available Condor marks them as reserved for the new job and notifies our tool by providing the reserved hostnames (this tool can run on the same machine as the Condor main service).
- Now our tool starts up and manages these hosts (connecting via Telnet, SSH). Condor shouldn't be involved in that part.
- Once our tool is complete it notifies Condor and Condor then marks the nodes as unreserved.

I really appreciate all the help I got on this mailing list and I'm sorry if my initial description was too vague, I hope the example above clears it up a bit.