[BUG] Module registration failed Error: Call timed out

[semenovnick]

Is this a bug in companion itself or a module?

• I believe this to be a bug in companion

Is there an existing issue for this?

• I have searched the existing issues

Describe the bug

The situation is quite similar to #2890. But in my case It began to manifest itself unpredictably on remote PC. I started trying to repeat it on my local PC.
After starting of connection/module, when on some reason process of connection/module is ended/corrupted (I made it from task manager). Companion runs in longtime loop of respawning the process with Module registration failed Error: Call timed out error, but connection status remains Connected for long period of time and after that ended with Unknown Stopped. I investigated the code and try to reproduce this error another way, it can be achieved only when forked process emmits exit. Seems like this listener do not hear 'exit' event

companion/companion/lib/Instance/Host.ts

Line 203 in e2eb7bf

child.monitor.on('exit', forceRestart)

And then Respawn forks new process, start runEntrypoint function of module and not recieving from companion registration approvemet, companion remains sending variablesChanged message instead of register...

Steps To Reproduce

1. Start any connection/module
2. End task via Task Manager or other meth

here is report from logs

report_companion.zip

Expected Behavior

After corrupting of process force restarting of connection/ or correct changing of status for manual restart by trigger / or correct communication with new respawn process.

Environment (please complete the following information)

- OS: Windoiws 10
- Browser: any
- Companion Version: catched in 3.4.3 and 3.5.1 also

Additional context

Thank you in advance

[semenovnick]

After initialization listener is removed for exit event in

companion/companion/lib/Instance/Host.ts

Line 183 in a7644ec

child.monitor.off('exit', forceRestart)

And that ilstener

companion/companion/lib/Instance/Host.ts

Line 203 in e2eb7bf

child.monitor.on('exit', forceRestart)

don't hear exit event... And then after unhandled error or sudden termination connection goes into that registration loop...

[Julusian]

That 'exit' handler should be fine to remove, It is still being handled inside 'respawn' with a restart being performed automatically.

The boundary of who is responsible for what around this restarting is a bit blurred, that respawn file was originally a library from npm. but the library looks abandoned and we needed to make some fixes to it, so it got adopted.
But this means that the logic we added on top of it to track restarts and some custom behaviour is not very clear/clean anymore.

[semenovnick]

Yes respawn restarts process successfully, but companion doesn't want to accept the new respawned process with new PID as e replacement of old crashed process and then when initialization starts and after:
https://github.com/bitfocus/companion-module-base/blob/58d91aaaf68ac3b3c654f9c605ffbcda61e2de3d/src/entrypoint.ts#L142
that Promise rejecting and new process is exit with code 11 than respawn restart process etc etc...
maybe Host after recieving exit Event should check and re-register new respawned process...