smartdispatch
smartdispatch copied to clipboard
SMART-Lab
An easy to use job launcher for supercomputers with PBS compatible job manager.
We faced a problem with the default module in GPU queues on Cedar. I added a CUDA module in the [Cedar config](https://github.com/bouthilx/smartdispatch/blob/cedar_config/smartdispatch/config/cedar.json) just it's done for others clusters ([helios example](https://github.com/SMART-Lab/smartdispatch/blob/master/smartdispatch/config/helios.json)),...
There are two log parts that ideally I would like to change: 1- The dir name inside SMART_DISPATCH_LOGS: I get names like this: 2017-10-27_10-25-49_**31_AFLW_gan_test_gan_coef**/ 2017-10-27_10-33-56_**31_AFLW_gan_test_gan_coef**/ 2017-10-27_12-18-48_**31_AFLW_gan_test_gan_coef_c_smart**/ 2017-10-27_12-26-25_**31_AFLW_gan_test_gan_coef_c_smart**/ So if I...
It is difficult to debug resuming while important process are taking place in the pbs script automatically built by SmartDispatch. Adding: - verbose to smart-dispatch/sd-launch-pbs command-line options - debugging prints...
The current autoresume feature is bonded to timeout. If the job is preempted by slurm, SmartDispatch won't reschedule itself. Slurm can be configured to reschedule preempted jobs, but it might...
