I know clusters exist to do this at the pipeline level, but I’d like to know if there’s anything on the roadmap that would allow this for jobs within pipelines.
My scenario: I mainly use spot instances for my agents and for the majority of our job steps this is a-okay. These are, however, easily preempted and for some steps I want to ensure they’re never preempted (a long running image build for example).
For these job steps, we have them target a smaller pool of standard, non-spot instances in our infrastructure.
The problem is: these agents will still pull jobs that may not need the standard VM.
We’re sorting this out a bit using priority settings and setting priority higher on the jobs that need standard VMs. But really what I’d like is a way to keep jobs in a pipeline that don’t specifically need a standard VM from even using these machines.
I’m pretty sure this is a feature request, but posting here just in case I’ve missed something that would allow me to do this now.
Thanks!