Rancher agent fails after reboot

Hi there.

I have a cluster built up on custom provider. It consists of 4 VM on a proxmox host.
The VM master uses CentOS 7, from the cloud init images. It works like a charm.
The problem is after a reboot of the VM the rancher agent fails to spin up with the error :

ERROR: https://kube.example.com/ping is not accessible (Failed to connect to kube.example.com port 443: Connection refused)
INFO: Arguments: --server https://kube.example.com --token REDACTED --no-register --only-write-certs

Indeed the agent can’t connect to port 443, as it’s not started :confused:
The cluster use external and internal IPs. I have the internal IP set up in the /etc/hosts file pointing to kube.example.com.

Do you have an idea of what’s going on ?

So it looks like I didn’t read all the documentation.

One have to mount the /var/lib/rancher folder to persist the configuration.

My bad.