ganrad / az-devops-agent-on-aks Goto Github PK

AZP_AGENT_NAME will be unique for each agent in a POOL.
If we are deploying two replicas only one will get created.The other pod will face an error "already has an active session exists"
Any possible to pass random name for AZP_AGENT_NAME?

I have a rbac cluster and I keep getting issues with my connections to use helm or kubectl in devops.

The agent running in AKS works like a charm, but it would be super convenient if I could have only one (or 0) running at all times, and a new pod would get created for every build, then killed after it's finished.

Any ideas on how to achieve that?

Team , I am unable to start docker demon, Please guide me how to start to use it in the runtime

Hi,

I just have question , what is the purpose of having ingress resource deployed for agent?

End users will just consume agent in their pipeline in Azuredevops portal and they will never hit the ingress. So had this question.

Once the cluster is migrated to 1.19 version which has containerd as the default runtime, the yaml deployment fails as it has volume mounts to /var/run/docker.sock(which will not be present in containerd runtime). Kindly help on the same on how to mount to /var/run/containerd/containerd.sock