We have a system where we have cryosparc workers installed on various GPU nodes and have masters installed on user workstations.
In my case I missed the upgrade to patch v3.2.0-20210413 and patched this weekend to v3.2.0-20210511. However the version of my master does not match that of the cluster. As you can imagine - only at specific is the cluster taken down for maintenance and upgrade. Is there a way to roll back the patch and apply a specific version?
Hi @Joe, there’s no formal way to roll back a patch without reinstalling. We have a workaround but it may cause some inconsistencies and we can’t guarantee that this won’t break the system:
I’m assuming your master is on 210413 and your worker is on 210511. If not do the reverse of each instruction:
Copy the folder contents of cryosparc_worker/cryosparc_compute/ into cryosparc_master/cryosparc_compute/
Replace the file cryosparc_master/patch with cryosparc_worker/patch
Run this command:
cryosparcm cli "complete_patch()"
Here’s what features that may break with this procedure, based on what changed between the two patches: