Automatic restart of an app (especially regarding websites) after it is down

[Daniel Jacob]

Why is there no automatic restart of an app (especially regarding websites) after it is down?

[girish]

@Daniel-Jacob app is already automatically restarted if it crashes. It's only not restarted, if it is running but "not responding" to health checks.

Is there a use case where one should restart an app when it's not responding? Other than, "let us try, why not" ?

[dsp76]

We experience it sometimes in the last days, where WordPress went down. At least in one case, when someone hammered our site and downloaded files from a plugin for whatever reason (DDOS?). Cloudron realized, the app is "not responding".

Unfortunately we realized it quite some time later. So an automatic restart attempt, when the app is not responding, would help in this scenario.

[girish]

@dsp76 thanks. Do you have any idea why a restart helps here? Maybe some processes got stuck? Usually, these are just either package bugs or app bugs. But if there are specific reproducible scenarios, I think we can investigate further .

I will move this to Feature Requests in any case.

[dsp76]

@girish we haven not finally found the reason. The "someone" mentioned actually is the internal IP Adress of the WordPress App. I only found one scenario in the web, when this might happen - if IPv6 is in use and the reverse proxy maps it with an internal IP. But IPv6 in our case is not enabled...

Its a specific wordpress plugin, which again and again will loaded with many of its files. Strange..

[dsp76]

@girish did you have seen other reports, where internal IP traffic in WordPress dev Image causes Wordpress to crash?

[girish]

@dsp76 I have not, no. So, you think (user installed) WordPress plugins are generating some internal traffic?

[dsp76]

Yes, randomly the WordPress Dev instance is crashing on the well powered system, while it was running on a native vServer stable before. The application log is not telling much about the reason - how can I see the Apache log for instance? I read https://docs.cloudron.io/packaging/cheat-sheet/#logging but have trouble understanding it, @girish

[girish]

@dsp76 the apache log should be in the main app logs - https://docs.cloudron.io/apps/#log-viewer

[dsp76]

@girish - the most likely reason my wordpress stops responding is this:

2024-05-19T22:56:54.000Z [Sun May 19 22:56:54.610795 2024] [mpm_prefork:error] [pid 1] AH00161: server reached MaxRequestWorkers setting, consider raising the MaxRequestWorkers setting
2

Is this something I can set?

[nebulon]

@dsp76 you should apply the same solution here as for other php apps like Nextcloud: https://forum.cloudron.io/post/83499

[dsp76]

Thanks! Increased it and will monitor and report, if it solved it.