Hey @wwsean08 - I found this linked from the blog post related to f5's Gitlab k8s configuration. Do you mind sharing how many executor replicas you configure to support 200 concurrent jobs and what requests/limits you give them? I'd love to get our executor resources a bit more dialed in.

Hey @robertr3, I unfortunately can't give you much more information as I left F5 a couple of years ago and don't remember all the details or have access to that information anymore. I think it was just a single executor that would spawn the 200 possible concurrent jobs on a bare metal kubernetes cluster that we ran in our data center, though I know we did have multiple executors as we had various "flavors" that were available with different limits in place based on teams needs. I think we also had a second runner paused at all times that way we could seamlessly upgrade where we would essentially upgrade the paused executor, then allow it to start picking up jobs, and stop the old executor from accepting new jobs so that it could complete the in progress jobs without causing issues for users. We had the burst capacity necessary to do this on our cluster, but I realize that may not be something that you have available to you.