Hi all, I'm launching a large number of external subworkflows, using reference launch plans. This is causing cluster load issues because each subworkflow gets it's own max_parallelism. Is there a way to force the workflows to share one resource limit, or alternatively to override the max parallelism for the subworkflows when i launch them with the reference launchplan?

I would love to understand what type of load issues are you seeing? Is it just that you have too many pods and cabinet keep up because you’re using reference launch plans. It is meant to be distinct and isolated units, though you can control the parallelism through the launch plan.

Ah it's just some rate limit for fetching container images IIUC

Yes this can happen. There are ways to Increase the container registry throttle limits. We did some tricks to deploy a base image everywhere Also if you are launching lots of small tasks - union has a cool way to reuse containers