This message was deleted.

Store your manifests in git and back up the manifests as text files rather than backing up the state of the cluster.

Yeah I agree with Trent. I use rancher backup chart to backup the rancher cluster itself to S3 on a schedule. For all other clusters, it's much more reliable and efficient to store YAMLs in Git and have something like ArgoCD that keeps the cluster in sync 🙂

ok, what is the restore process? Will it be: • restore Rancher states on new clusters • Define a playbook, so that it can re-deploy everything, even if the ArgoCD is crashed?

For rancher: Use rancher backup Create a new cluster Install backup operator Restore rancher tar file For k8s: Create new cluster Install Argo and configure it to look at gitrepo Argo deploys manifests With fleet you can skip step 2 Cattle not pets. Kubernetes is designed to be ephemeral and declared. Trying to use kubernetes in an imperative and stateful manner is an anti pattern.

I use this guide: https://ranchermanager.docs.rancher.com/pages-for-subheaders/backup-restore-and-disaster-recovery The easiest way in my experience is to just create a new one. That way, you can update the underlying K8 version as well as the rancher version itself while you are at it. It's also 0 downtime. Once you backup using the above doc, you can follow this: https://ranchermanager.docs.rancher.com/how-to-guides/new-user-guides/backup-restore-and-disaster-recovery/migrate-rancher-to-new-cluster After the new cluster is operational, I just update the dns entry for the old one to point to the new one and everything magically works 🙂

will it works? if I follow this way: • restore Rancher state in another new cluster, just for management cluster • Re-deploy all manifest I did on downstream clusters (assuming I have a playbook that does) I have a bunch of Helm charts for my applications....FYI

So for me, If I restore the state on the new cluster, everything from the downstream clusters is just there. You don't have to redeploy unless you want to change something 🙂