Can't reach other pods or external from a second node pod (Tailscale?)

[dMARLAN]

I'm working on getting a multi-node k3s cluster set up (we've been running single node for awhile now)

I think I'm almost there, I can get an agent to join the cluster, pods are scheduled/started on the agent when the server applies new deployments...

but the pods can't reach each other as far as I can tell. I tried exec -it into a pod, but they currently don't have curl installed, and apt-get update hangs.

root@ubuntu-deployment-6b8c459959-hpb9l:/# apt-get update
0% [Connecting to archive.ubuntu.com] [Connecting to security.ubuntu.com]
(stuck)

I'll try with an image that already has curl installed shortly, although I expect it won't be able to reach the pods either.

I launched the server with:
(I am using Tailscale, which is why I added --advertise-address which fixed other issues, although I assume further configuration is required to make this work)

sudo k3s server --advertise-address=100.94.157.87 --write-kubeconfig-mode 644

and launched the agent with:

sudo k3s agent -t foo::server:bar -s https://100.94.157.87:6443

[dMARLAN]

The pods are fine on the same node, just not if they are on my agent node.

I'm viewing this: #10078 and going to try playing with my Tailscale ACLs

[dMARLAN]

I don't think the issue is ACLs, it's set to allow all currently.

[dMARLAN]

It was because I was running my k3s server on a VirtualBox VM and the default Network Adapter is NAT but it needs to be set to Bridged

<internal screaming>