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

Re: [HTCondor-users] Simple use case



On 12/16/2013 7:39 PM, David Katz wrote:
Can Condor submit N jobs in parallel to a specific set of nodes; I want them to run once and only once on each node.

Thanks!

David Katz| IAG, TIBCO Spotfire


So lets says you have three nodes: nodeA.com, nodeB.com, nodeC.com

Could be as simple as a submit file that looks like so:

executable = foo
requirements = Machine == "nodeA.com"
queue
requirements = Machine == "nodeB.com"
queue
requirements = Machine == "nodeC.com"
queue

Or lets say you had 50 nodes named node0.com, node1.com, ... node49.com:

executable = foo
requirements = Machine == "node$(Process).com"
queue 50

Or lets say you had 50 nodes named node1.com, node2.com, ... node50.com:

executable = foo
requirements = Machine == "node$$([1 + $(Process)]).com"
queue 50


For all of the above, the jobs will start on each machine once the machine has a matching available slot. For instance, if 48 out of 50 nodes are available to run a job at time T, then 48 jobs will start simultaneously and the last two jobs will start whenever those nodes become available to you (i.e. they are not in use by a higher priority user). Hopefully this is what you were after. If not, please provide more details.

Also, this URL may be helpful to you:
  http://research.cs.wisc.edu/htcondor/quick-start.html

regards,
Todd