Add worker nodes

Install, Configure and Update
1

Greetings!

We recently added new nodes to our cluster. They are working fine, but Cryosparc jobs do not get scheduled on them.

Do we need to update the Cryosparc configuration to make it recognize the new workers? If so, how can we do this without otherwise changing the Cryosparc configuration?

Thanks!

2

In some cryoSPARC instances that use an external (to cryoSPARC) workload manager, the addition of worker nodes may be handled exclusively by that external workload manager’s configuration.
If that’s not the case, the necessary steps are described in this section of the guide.
That section assumes that the software is already available, either by separate installation or by sharing, on the worker nodes to be added.
Please let us know if you have any questions.

3

Thanks!

Cryosparc is submitting the jobs to slurm. Is there way to check the configuration to confirm that the setup is exclusively using the external workload manager? It has been quite a while since we initially set this up.

4

The are the get_scheduler_lanes() and get_scheduler_targets() cryosparcm cli functions.

You can use cryosparcm cluster dump <name> to recreate the configuration files that were used when the cluster was connected to your cryoSPARC instance earlier.

In a scenario where the new workers have similar capabilities to the old ones and are intended to add additional computational capacity, you could:

  1. Ensure the worker software tree is available on the new workers (via installation or, more likely, a shared filesystem) at the same path as on the old workers.
  2. Integrate the the new worker nodes to the slurm partition that already includes the older cryoSPARC workers. This process should not entail any reconfiguration inside cryoSPARC.