k3s: [release-1.21] Panic checking name of uninitialized etcd member

Environmental Info: K3s Version: k3s version v1.21.7+k3s1 (ac705709) go version go1.16.10

Node(s) CPU architecture, OS, and Version: Raspberry Pi Linux k8smaster1 5.10.83-1-MANJARO-ARM-RPI #1 SMP PREEMPT Tue Dec 7 08:43:25 CST 2021 aarch64 GNU/Linux (They all have the same image, but with different static IPs and hostnames: k8smaster0, k8smaster1, k8sworker0, k8sworker1.

Cluster Configuration: 1 server, 2 agents The server had --cluster-init passed to it, and it has etcd.

Describe the bug: I attempted to add a second master and got a strange error. The first master was given --cluster-init, so etcd was in its roles. Here’s the end of the log if I try to start it manually on the command line with k3s server:

INFO[2021-12-17T19:52:17.353594206-05:00] Containerd is now running                    
INFO[2021-12-17T19:52:17.384384680-05:00] Connecting to proxy                           url="wss://127.0.0.1:6443/v1-k3s/connect"
ERRO[2021-12-17T19:52:17.394193344-05:00] Failed to connect to proxy                    error="websocket: bad handshake"
ERRO[2021-12-17T19:52:17.394287287-05:00] Remotedialer proxy error                      error="websocket: bad handshake"
panic: runtime error: slice bounds out of range [:-1]

goroutine 232 [running]:
github.com/rancher/k3s/pkg/etcd.(*ETCD).join(0x4000e2f770, 0x4b02318, 0x4001151c80, 0x4001342000, 0x0, 0x0)
	/go/src/github.com/rancher/k3s/pkg/etcd/etcd.go:350 +0xfd8
github.com/rancher/k3s/pkg/etcd.(*ETCD).Start.func1(0x4000e2f770, 0x4b02318, 0x4001151c80, 0x4001342000)
	/go/src/github.com/rancher/k3s/pkg/etcd/etcd.go:307 +0x58
created by github.com/rancher/k3s/pkg/etcd.(*ETCD).Start
	/go/src/github.com/rancher/k3s/pkg/etcd/etcd.go:305 +0xe0

It may be worth noting that I was able to join 2 worker nodes without a problem.