Skip to content

minor osg workflow change to allow recovery of in-progress data

Turns on file transfers on eviction.

On checkpoints and evictions, job sandboxes are copied to /var/lib/spool. This can be a hassle to find but as long as the job is in the queue, you can find the copy of your sandbox with condor_evicted_files <jobid>.

So, if you want to get hands on data from an in-progress job, you could do:

$  condor_vacate_job <jobid>
$  condor_evicted_files <jobid>

Which will evict your job and you'll be able to find the most recent data. You can also skip the vacate step and get the most recent checkpoint before then (which will probably work out to be the same thing anyway).

This MR also updates the directives for running things like bayeswave_post and megaplot in the local pool.

Merge request reports