SSD parameter in cluster template

Install, Configure and Update
1

I have CryoSPARC running on a cluster. Everything works fine but there is one problem that I am trying to solve. That is, how to prevent more than one CS job that uses SSD from landing on the same node, because our SSDs are not large enough for more than one job. I know how to do it via a resource request but I do not know is when to set the request. I need to know whether a job will use SSD so that I can set the proper request value in the template.

The job.json file has the following, as an example:

{
  "slots": {
    "CPU": 8,
    "GPU": 4,
    "RAM": 2
  },
  "fixed": {
    "SSD": false
  }
}

Is there any way to get the “SSD” attribute exported as a variable to the template script? So I could do something like the following in the template submission script?

#$ -l ssd={{ use_ssd }}
...

Can it be added to template_args in cryosparc_master/cryosparc_command/command_core/__init__.py?

1 Like
2

Thanks for your suggestion @glennpj. We have taken note.
A potential caveat: such parameter would prevent host sharing between two related jobs that otherwise could work with the same cached data.