Cluster FAQ

Agent installation

Q: What if I install the same agent manifest on multiple clusters?

Installing the same agent manifest on more than one cluster can result in the misconfiguration of both clusters and unexpected behaviour in the Conductor console. To avoid this issue there are guardrails in place that will put the agent pods in CrashLoopBackoff status when the manifest has already been installed on another cluster. The agent will log messages similar to the following:

level=error msg=problems authenticating with upstream system, err: problems authenticating on initial pinger: rpc error: code = AlreadyExists desc = agent already deployed on different cluster: agent been deployed on multiple clusters with same cluster ID 

level=error msg=this agent has already been registered for another cluster. If you have migrated to a new cluster or have recreated the agent deployment, we must wait for the old agent id to expire before we can successfully connect, please be patient

Q: Does Conductor support ARM-based clusters?

Yes, but some configuration might be required to ensure successful Dapr control plane and Conductor agent deployments. See Install Conductor on ARM Cluster