A simple quality-of-life improvement would be if in the beginning of the run print the name of the worker computer running the job to the event log so that it is easier to troubleshoot problems with individual compute nodes on a shared resource.
As it is now, only the name of the lane is printed. (Or perhaps there is already a way to deduce which worker node was used?)
Hi Daniel,
At least on our setup, I note that in addition to the lane, the worker hostname is reported in the streamlog (at the top of the Resources allocated section) and job.log. We don’t utilise an external scheduler at the moment, but in my previous institute, I remember this information being similarly captured under cluster-lane conditions. Do you see different?
Cheers,
Yang
Perhaps you had an “echo $hostname” statement in the job template
