Hello everybody.

[autumn moat]

I don't use gitlab's runner operator and use the gitlab-runner helm chart instead and add dagger's runner to the helm values for the gitlab-runner chart. Looks like the docs have been updated since I set it up. operator vs helm depends on your comfort level, I suppose. I have the dagger-engine daemonset with a nodeselector that is tied to the gitlab-runner build pods which allows for more control over the nodes that the CI runs on.
for my setup, the runner from the helm chart runs in the gitlab-runner namespace and accesses dagger-engine via the buildkitd.sock on the host, no RBAC needed