Messenger process managers

[tvlooy]

Add information about mysql timeouts, supervisor FATAL, systemd user services.

Or, how should you handle this MySQL timeout issue, supervisor restart (FATAL) issues? How do you people do this? What does the component author thinks about this?

I myself am using with systemd user services in production for a while and it's running okay.

[tvlooy]

did some polishing. This PR is ready

[vicla31]

FI, [Unit] section of service allow "Requires=" which could be set to prevent failure from dependency not meet a startup

[OskarStark]

@tvlooy can you please retarget to 5.1? Thanks

[tvlooy]

FI, [Unit] section of service allow "Requires=" which could be set to prevent failure from dependency not meet a startup

What would you suggest as a Requires? After=network.target rabbitmq.service something like that? I don't do that for my services atm

[tvlooy]

@OskarStark thanks for the review!

[tvlooy]

No interest in merging this?

[icanhazstring]

Please merge this, as I was looking days for a solution on supervisor. Thanks to @tvlooy we managed to implement the wrapper script which added the needed delay between restarts. As the PR for supervisor does not provide this.
Supervisor/supervisor#487

So a merge would be nice for the world to know :)

[wouterj]

Hi @tvlooy! Thanks a lot for sharing this knowledge by contributing to the docs. Having this expert information in the docs is great, but - as you noticed - it's also hard to review/merge as we aren't necessarily expert on the topic ourselves :)

As a side note: this type of information is often great for a personal blog post. In that way, you can still help the community by sharing your knowledge and it's often a lot "lighter" (having the official stamp on the docs means we have to be very careful on what to recommend).

Nonetheless, I think having a systemd example in the documentation is great and looking at the comments in this PR, the community is looking for a way to avoid FATAL states.

Reading the Supervisor issue referenced by @icanhazstring, I found out about Supervisor's BACKOFF strategy: Supervisor/supervisor#487 (comment) I've decided to document that instead of your bash helper script, as it feels more official. Let me know if you think this doesn't work for what you intended to solve :)