FATA[0000] ssh dial error: ssh: handshake failed: EOF

[kvelev]

Hello, I'm trying to deploy upterm using the helm chart available in this repository and I've run into the following error when I attempt to establish a session.

upterm host --server ssh://upterm-domain.ltd:22 -- bash

FATA[0000] ssh dial error: ssh: handshake failed: EOF   

Upon turning on the debug mode on I can see that the pod is constantly trying to connect to some internal IP:

time="2024-10-30T09:38:32Z" level=debug msg="connection establishing failed" addr="10.0.0.107:53998" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"
time="2024-10-30T09:38:42Z" level=debug msg="connection establishing failed" addr="10.0.0.107:46476" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"
time="2024-10-30T09:38:52Z" level=debug msg="connection establishing failed" addr="10.0.0.107:45770" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"
time="2024-10-30T09:38:52Z" level=debug msg="connection establishing failed" addr="10.0.0.107:45762" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"
time="2024-10-30T09:39:02Z" level=debug msg="connection establishing failed" addr="10.0.0.107:39428" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"
time="2024-10-30T09:39:12Z" level=debug msg="connection establishing failed" addr="10.0.0.107:58034" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"
time="2024-10-30T09:39:12Z" level=debug msg="connection establishing failed" addr="10.0.0.107:58024" app=uptermd com=ssh-proxy error=EOF network=mem network-opt="[]" node-addr="XX.XX.XX.XX:22" ssh-addr="10.0.0.168:22"

I have only 1 upterm pod - 10.0.0.168 and I have also checked - there's no resource with the IP of the addr - 10.0.0.107 - what is this IP? What is the addr value representing in this case? Seems to me the pod is trying to connect to a non-existing resource.

Furthermore, I've noticed that in the helm deployment the node IP is actually the pod IP - why is that?

Also, if I want to use a pub/private key pair that I store in the secret that is used to mount the volume - https://github.com/owenthereal/upterm/blob/master/charts/uptermd/templates/secret.yaml - what should the name of the key be? id_rsa? key?

I'd appreciate some help.

Thanks.

[mxschmitt]

I was running into ssh: handshake failed: EOF error as well. When starting an upterm server, you need to configure @cert-authority in ~/.ssh/known_hosts:

ssh-keyscan uptermd.upterm.dev 2>/dev/null | grep -v '^#' | awk '{print "@cert-authority * " $2 " " $3}' >> ~/.ssh/known_hosts

This gets done automatically when creating a session via upterm and it asks for accepting the fingerprint or alternatively via the command above for non-interactive sessions. In action-upterm this gets done here.