Issue Unresponsive issues with server

[cguanaja]

Server operating system version

Ubuntu 20.04.6 LTS

Plesk version and microupdate number

Plesk Obsidian 18.0.51 Update #1

Hi community,

I've been having some ongoing issues with below two recurring messages:
- The threshold of Apache & PHP-FPM memory usage has been exceeded
- The swap usage threshold has been exceeded

I've also checked below URL's:
- Issue - The threshold of Apache & PHP-FPM memory usage has been exceeded
- Issue - Apache & PHP-FPM Memory Usage Growing Exponentially - tail commands give no output on affected server on my end
- Issue - Swap Usage Threshold Exceeded

I've left atop service running a while back which may have been causing the web host to exponentially be unresponsive, I was not even able to login via SSH and being forced to restart the AWS instance. Upon checking the top command now I see there are services that come every now and then from each domain with admin user for each domain and in relation to a "php-fpm: pool domain-name" command, and these a lot of memory.

I'm leaving the atop service stopped - is this how it should be left so it won't cause any memory/swap usage hiccups? Screenshot
What else can I check so this "exponential" memory/swap usage does not cause server unresponsiveness?

I look forward to your input.

Thanks!

 

[Peter Debik]

One ore more websites are creating long running PHP-FPM processes. Some basic configurations can already help limiting the CPU use: Decrease the PHP timeout values of your website configurations to something low like 10 seconds and limit the number of concurrent children to a low number like 10.

Make sure to look into access_ssl_log and error_log files of the affected websites that consume the most CPU time to find out more on the reason. Often bad bots are hitting a site. Enable the Fail2Ban BadBot jail and maybe implement additional bad bot protection. Some guides:

How to Secure Nginx Against Malicious Bots

Protective measures for a server are very important and there are several ways to protect your websites and apps from malicious bots. We’ll be looking at different bots and how they operate, and how you can use Plesk’s security measures to secure Nginx against malicious bots. Malicious Bot Types...

www.plesk.com

Question - Blocking bad bots

The server seems to be mainly servicing https bots while its running, and it annoyed me. So I wrote some nginx code to stop it (inserted via the PLESK interface into the nginx configurations box): if ($http_user_agent ~...

talk.plesk.com

and others.

 

[cguanaja]

Thanks for your input @Peter Debik - changed both max_execution_time and max_input_time - set both to default values for PHP.

I saw below error before/after doing changes - and before I go through the URL provided, what errors might be causing this on the Plesk host?

New configuration files for the Apache web server were not created due to the errors in configuration templates: nginx: [emerg] bind() to [2600:1f14:fc4:1600:1663:41fc:f969:d3a3]:443 failed (99: Cannot assign requested address) nginx: configuration file /etc/nginx/nginx.conf test failed. To fix the issue, follow the instructions described in KB: https://support.plesk.com/hc/articles/213933865

Many thanks.

 

[Peter Debik]

Please check this KB article:

https://support.plesk.com/hc/en-us/articles/12377844474903-Nginx-on-Plesk-server-fails-to-start-99-Cannot-assign-requested-address

 

[cguanaja]

Thanks @Peter Debik - I checked but I don't get the IP address marked as broken so I don't get the option to repair it - strange.

Refreshed browser and error on dashboard is still there.

Thanks.

 

[cguanaja]

Any further input here please? I've had to reload instances twice so far.

I get these messages:
The swap usage status is critical!
The threshold of Apache & PHP-FPM memory usage has been exceeded

I loose connectivity to the host even via SSH, and every time I reload the instance it goes back to normal which is not optimal.

What might be causing this?

Thanks.

 

[manni]

Hi,

I had this situation as well, already a few months ago. May I ask you: Do you use nginx as layer 1/proxy and apache then for processing PHP?
If so, I'd suggest to try it without nginx. In our case nginx+apache combination caused this issue.

best,
manni

 

[cguanaja]

Hi @manni - thanks so much for your reply. I've checked and all Common Apache settings are set to default, and the following nginx settings are enabled:
Proxy mode and Smart static files processing.

This applies to all domains - are you saying that disabling "proxy mode" under nginx settings for all domains can help?

Many thanks.