Closed glsf91 closed 3 years ago
Correct. These services are all based around a "fail fast" model that will stop the entire process on failure. When running in Docker, this results in a container restart and allows the app to start from scratch, i.e., this is expected behavior because it is built around a container / service manager monitoring externally.
As for the error, your HomeGraph API file seems like it might be missing, bad, or the HomeGraph API Google setup hasn't been completed.
I'am not very familar with Docker. So this worked as designed. With systemd I expect I can do the same.
Correct, as designed. I basically side stepped a lot of error / reconnect logic by saying "on any error just log and restart". Its not he most efficient, but it keeps things clean as this is ONE of like 7 or 8 little bridges like this I manage, and they all run the same way and have the same error handling approach with vastly different client implementations.
In Windows, you could run as a service to get the same restart functionality.
In Linux you'd have to setup some initd / whatever the current system process management framework is this week to do the same.... but thats why I highly recommend the Docker usage as its way easier.
I saw the programs crashes when something is going wrong with communications to google:
I think it should never crash but abort the current action and continue running.