On 10/29/2012 10:15 AM, Sarah Williams wrote:
Hello all, In condor I see sometimes that, if a node has a problem and can't complete a job, the system will re-queue it and retry it on another node. However many of the jobs we get don't support that, and the jobs fail anyway in addition to confusing upstream monitoring. Is there a way to tell condor to never retry a job? --Sarah
Condor will keep track of how many times your job has been started in the job's NumJobStarts attribute.
If you only want a job to start once you can add the following to your submit file -
requirements = NumJobStarts == 0 Best, matt