When deploy fails, some child processes seem to persist beyond the parent

[wmertens]

Most obvious when I'm doing --auto-rollback false. The deploy fails and returns to prompt, and after that I get the message 👀 ❌ [wait] [ERROR] Error waiting for activation: Timeout elapsed for confirmation.

[wmertens]

I have further information on this: The process is activate-rs.

It might be because of my ssh configuration that uses a persistent connection with an auto master?

In any case, I'd expect that when I press Ctrl-C, the activate-rs SSH command also gets killed, and that when the deploy fails, the activate-rs is awaited.