When I configure Github-hosted Runners, there is a "max parallelism" text field. Depending on who I ask and which help page I read, I can interpret this in two different ways:
- The runner will spin up one "machine" (VM) and up to N jobs can run on this VM and share the resources. If I want more VMs to be spun up in parallel, because I have a very active repository, I should reduce this number.
- The GitHub infrastructure will spin up one "machine" (VM) per job (workflow) that gets started, and at most N VMs of this type will be active at once. If I want more VMs to run in parallel, because I have a very active repository, I should increase this number.
Which of these two interpretations is correct, or is there a third one?
