Hey 
I investigated today if using the “Gitlab Runner Operator” is fitting for my usecase. In short: is it capable of managing runners on a multi cluster? The docu is in general a little brief.
We have a self managed Gitlab instance (docker using harbourmaster) on a local machine. We have additional machines with a k3s cluster each. These k3s clusters used for the runners. All of them are using the docker executor. Additionally we have an aws eks cluster, where we also provide runners, using the kubernetes executor, mainly to have docker in docker capabilities.
Our current runner definition/deployment setup is a repo, where the first layer of subdirectories is split up per “location”, so one per k3s cluster and the eks cluster. The next layer of subdirectories is each of the “location” containing folders with a helm chart for each runner “class”. The helm charts are deployed by argoCD, that is running in our EKS cluster as well.
I want to rebuild the runner management to enable us more flexibility, like a central management of hardware resources a certain runner “class” is assigned to. Then I would be enabled to change a certain runner class and not do it manually in each chart.
I now had the idea, that this might be a usecase for the Gitlab Runner Operator. So: Is it capable of managing runners in different clusters? As said in the beginning, I couldn’t find details on that in the docu, also not by using google or searching this forum.
Otherwise I will try to achieve that by adding another helm chart layer and a configmap that fits my needs