Hi 
Since import job on master consume a lot of resources, except using queue system like slurm, is there any other way to make it running on worker node?
If you have one or more worker nodes that were connected to your CryoSPARC instance with the
cryosparcw connect command, you may try setting
export CRYOSPARC_DISABLE_IMPORT_ON_MASTER=true
(details).
Great !
I used to think that only works on cluster.