Error assigning CPUs to worker


we are having issues with assigning CPUs to a worker. We have 96 CPUs and 8 GPUs, and we want to assign 24 CPUs and 4 GPUs to one worker. We found the following error:

cryosparcw connect --master -n -worker bremlive-worker --port 39000 --ssdpath /home/cryosparcuser/scratch/cryosparc_cache --newlane --lane bremlive --cpus 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24 --gpus 0,1,2,3

usage: [-h] [–worker WORKER] [–master MASTER] [–port PORT]
[–sshstr SSHSTR] [–update] [–nogpu] [–gpus GPUS]
[–nossd] [–ssdpath SSDPATH] [–ssdquota SSDQUOTA]
[–ssdreserve SSDRESERVE] [–lane LANE] [–newlane]
[–rams RAMS] [–cpus CPUS] [–monitor_port MONITOR_PORT] error: argument --cpus: invalid int value: ‘0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24’

Without the CPU part, everything works well. We can create a worker with 4 GPUs, but we cannot modify the CPUs or the RAM. Are we doing something wrong?

Thanks for your help!

Hi Jorge,

I’m not sure if it’s the intended behaviour, but having added the worker node, I believe you can modify the --cpus and --rams parameters using the --update flag. For example, --cpus 24 and --rams 8 will update it to use 24 CPU slots and 64GB worth of memory.

cryosparcw connect --master --worker bremlive-worker --port 39000 --cpus 24 --rams 8 --update


Yes. The --cpu parameter should be specified with a single integer. The guide has been updated accordingly.

Hi @wtempel,

I think the underlying issue may be that the --cpus and --rams parameters seem to be ignored when adding a new worker node, either to a new lane or an existing lane. They do take subsequently when applied with the --update flag. --gpus is respected in both scenarios. It seems like a bug to me.


Confirmed. We have taken note.
Thank you for reporting the problem @jplopez @leetleyang .