Not sure if interesting due to my env setting - running rke2 cli commands from inside a container for UT - but I get this occasional panic at the final stages of

cluster-reset

. If it'll be useful I can open an issue.

2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Connected to etcd v3.5.21 - datastore using 16384 of 20480 bytes"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Defragmenting etcd database"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Connected to etcd v3.5.21 - datastore using 12288 of 20480 bytes"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Defragmenting etcd database"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Datastore using 12288 of 20480 bytes after defragment"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Datastore using 12288 of 20480 bytes after defragment"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Connection to etcd is ready"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="ETCD server is now running"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="rke2 is up and running"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Saving cluster bootstrap data to datastore"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Bootstrap key locked for initial create"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="certificate CN=etcd-peer signed by CN=etcd-peer-ca@1760959689: notBefore=2025-10-20 11:28:09 +0000 UTC notAfter=2026-10-20 11:28:50 +0000 UTC"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="certificate CN=etcd-server signed by CN=etcd-server-ca@1760959689: notBefore=2025-10-20 11:28:09 +0000 UTC notAfter=2026-10-20 11:28:50 +0000 UTC"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Shutting down kubelet and etcd"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=error msg="Kubelet exited: signal: killed"
2025-10-20T11:28:50+00:00 >> time="2025-10-20T11:28:50Z" level=info msg="Bootstrap key lock is held"
2025-10-20T11:28:50+00:00 >> {"level":"warn","ts":"2025-10-20T11:28:50.180515Z","logger":"etcd-client","caller":"v3@v3.5.21-k3s1/retry_interceptor.go:63","msg":"retrying of unary invoker failed","target":"<etcd-endpoints://0xc00125a3c0/127.0.0.1:2379>","attempt":0,"error":"rpc error: code = Unavailable desc = error reading from server: EOF"}
2025-10-20T11:28:50+00:00 >> panic: rpc error: code = Unavailable desc = error reading from server: EOF
2025-10-20T11:28:50+00:00 >> 
2025-10-20T11:28:50+00:00 >> goroutine 629 [running]:
2025-10-20T11:28:50+00:00 >> <http://github.com/k3s-io/k3s/pkg/cluster.(*Cluster).Start.func2()|github.com/k3s-io/k3s/pkg/cluster.(*Cluster).Start.func2()>
2025-10-20T11:28:50+00:00 >> 	/go/pkg/mod/github.com/k3s-io/k3s@v1.33.2-0.20250616230217-4256d5813c74/pkg/cluster/cluster.go:79 +0x165
2025-10-20T11:28:50+00:00 >> created by <http://github.com/k3s-io/k3s/pkg/cluster.(*Cluster).Start|github.com/k3s-io/k3s/pkg/cluster.(*Cluster).Start> in goroutine 1
2025-10-20T11:28:50+00:00 >> 	/go/pkg/mod/github.com/k3s-io/k3s@v1.33.2-0.20250616230217-4256d5813c74/pkg/cluster/cluster.go:72 +0x19b

Wait, what are you doing?

The entire cluster (single node) is running inside a container using the cli alone and not through systemctl. I understand this isn't the way to go, this is only meant for some specific UT scenarios, yet as this panic isn't persistent it might suggest some race.

Is the main rke2 server process also running when you run this command?

no

you might try on a newer release? That one’s a few months old, and that bit of code has already been refactored to no longer panic if the save fails.

sure, thank you