Is it at all possible to restore an RKE2 downstrea...
# general
s
Is it at all possible to restore an RKE2 downstream cluster from etcd snapshots in Rancher, if the cluster is completely lost? If I enable storing the etcd snapshots in S3, then the snapshots are persisted. But I still need the RKE2 server token to restore the snapshot, and I don't think that is available anywhere if the downstream cluster gets deleted. Of course I can backup the downstream cluster's hosts, but it feels pretty wasteful to backup all that when I just need the server token… 😕 Is this case simply not handled by Rancher, or am I missing something?
c
yes. Rancher knows the token, if you provisioned the cluster via Rancher. As long as you have an etcd backup still on s3 you can recover from loss of all nodes.
s
Oh, it does!? Great. Any idea where? I've been looking through the
cluster
objects to no avail… 😕
Because the full token is required, right? It seems like the short credentials part of the token can be found in the secret
*-rke-state
in the
fleet-default
namespace. But I guess I also need the CA hash? 🤔
No, it seems like the short token is sufficient to decrypt the snapshot and then the CA hash is extracted from the snapshot (I'm guessing). Thanks for pointing me in the right direction!