SSH connection over Mycelium network becomes unstable after some time

[Mahmoud-Emad]

Description:

When using the Mycelium network to establish an SSH connection between two machines, the connection drops unexpectedly after some time.
Once the connection is lost, restarting the Mycelium process (mycelium binary) restores connectivity, but the issue reoccurs periodically.

This happens consistently and seems unrelated to SSH client/server configuration.

Steps to Reproduce:

1. Start Mycelium on two machines and connect them via public or private nodes.
2. Establish an SSH session between the two machines over the Mycelium network.
3. Keep the SSH session active for a while (e.g., several minutes).
4. Observe that the SSH connection becomes unresponsive or disconnects.
5. Restarting the mycelium process temporarily resolves the issue.

Environment:

Mycelium Version:

• 0.6.1 (observed by me)
• 0.6.2 (latest version — same issue observed by @AbdelrahmanElawady )

Platform: MacOS

[scottyeager]

Let's do the following to collect info that will be helpful for debugging this:

• Enable debug logging for local mycelium (add -d in the systemd service)
• At the time of connection difficulty, collect outputs from the following commands:
  • mycelium routes selected
  • mycelium routes fallback
  • mycelium routes queried
  • mycelium routes no-route
• Collect logs, ideally focused around the time the issue begins
• Note the destination Mycelium IP

Here's a bash script to collect all of the outputs into a file:

for cmd in selected fallback queried no-route; do
  echo "--- mycelium routes $cmd ---" >> routes_output;
  mycelium routes "$cmd" >> routes_output;   
  echo "" >> routes_output;
done

Attach the routes_output and the logs to this issue.

[mik-tf]

Thanks Mahmoud and Scott. Good idea.

So tbh I am lazy and when I end up having to restart mycelium it's often in a code-frenzy session so I am too impatient to save the logs. Thus here's a little trick:

I’ve set up a small helper script so that when Mycelium gets into a bad state and I restart the systemd service, I don’t lose the useful debugging info (routes + logs).

What I’m doing now

• Script location: ~/.mycelium-logs/myceliumr.sh
• Alias: alias myceliumr="$HOME/.mycelium-logs/myceliumr.sh"
• Usage: when SSH over Mycelium hangs, I run:

myceliumr [mycelium-ip-address]

What the script does

1. Ensures ~/.mycelium-logs/ exists.

2. Creates a new log file named with a UTC timestamp, e.g.
   ~/.mycelium-logs/2025-12-11T20-05-56Z.log.

3. If an address is provided as an argument, prepends "Mycelium address:

   " to the file.

4. Appends the current routing state:

   • mycelium routes selected
   • mycelium routes fallback
   • mycelium routes queried
   • mycelium routes no-route

5. Appends recent journal logs for the service:

   sudo journalctl -u mycelium.service -n 1000

6. Finally runs:

   sudo systemctl restart mycelium

So each time the issue occurs, I can run myceliumr instead of systemctl restart mycelium directly, and I get a timestamped bundle with both mycelium routes … output and the relevant journalctl logs from right before the restart. I’ve tested this flow locally and confirmed that the files are created and the service restarts successfully.

How to set this up

1. Create the script and directory:

mkdir -p ~/.mycelium-logs
cat > ~/.mycelium-logs/myceliumr.sh << 'EOF'
#!/usr/bin/env bash
D="$HOME/.mycelium-logs"; mkdir -p "$D"
F="$D/$(date -u +'%Y-%m-%dT%H-%M-%SZ').log"
if [ -n "$1" ]; then echo "Mycelium address: $1" > "$F"; fi
for c in selected fallback queried no-route; do
  { echo "--- mycelium routes $c ---"; mycelium routes "$c"; echo; } >>"$F"
done
sudo journalctl -u mycelium.service -n 1000 >>"$F"
sudo systemctl restart mycelium
echo "Saved debug info to $F"
EOF
chmod +x ~/.mycelium-logs/myceliumr.sh

2. Add a convenient alias:

   • Bash / Zsh (e.g. in ~/.bashrc or ~/.zshrc):

     alias myceliumr="$HOME/.mycelium-logs/myceliumr.sh"

   • Fish (in ~/.config/fish/config.fish):

     alias myceliumr "$HOME/.mycelium-logs/myceliumr.sh"

3. Reload your shell config or open a new shell, then run myceliumr when the issue occurs:

   • Bash / Zsh: source ~/.bashrc or source ~/.zshrc
   • Fish: source ~/.config/fish/config.fish

This will leave you with timestamped log files under ~/.mycelium-logs/ that you can attach to this issue.

---

So I will now use this and will share those log files soon.

I can say it happens maybe 3-4 times a week that I have to restart mycelium. Is it because of this bug exactly? Not sure but we will find out with the log files.