Restarting a process is not working on some termination conditions.

[offbyone]

Overmind: v2.5.1
OS: macOS Sequoia 15.1.1

I have a Procfile with a command that connects to a GitHub codespace for port forwarding:

port-forward: gh cs --codespace $(script/get-codespace-id) ports forward 80:3080 3035:3035
ngrok: ...

I have configured that service to be allowed to die and to restart:

OVERMIND_CAN_DIE=port-forward,shell,console
OVERMIND_AUTO_RESTART=port-forward,shell,console

When I have a network interruption, though, the process dies, and is not restarted:

$ overmind status
PROCESS            PID       STATUS
port-forward       12497     dead
ngrok              84235     running

Connecting to the port-forward process shows this:

Forwarding ports: remote 3035 <=> local 3035
Forwarding ports: remote 80 <=> local 3080
error connecting to tunnel: connect to forwarded port failed: error connecting to forwarded port: failed to open streaming channel: failed to open port forward channel: failed to open channel: read tcp 192.168.5.148:64831->20.69.79.91:443: read: can't assign requested address

... and that's it. It doesn't restart, which is what I expect given the configuration.

[DarthSim]

Hey @offbyone!

You should not use OVERMIND_CAN_DIE in this situation.

Processes in the OVERMIND_CAN_DIE list can die without any consequences. Their termination won't cause the termination of the rest of the processes, nor will Overmind restart them.

OVERMIND_AUTO_RESTART == "if it dies, try to revive it"
OVERMIND_CAN_DIE == "if it dies, let it be dead"

[offbyone]

TIL! Thanks!