So I have been trying to go through an issue, where after a certain spike in PHP operation which exceeds in Either IOPS, CPU/RAM threshold - it brings not only the site to a complete stalemate but all PHP Operations goes into Sleep state.
This happens both on Nginx + php-FPM or Apache + PHP-FPM
In certain cases
- PHP FPM restart does help
- PHP FPM restart does not help and need to manually kill all PHP processes.
I'm aware that this must be due to a load spikes or possibly a pm based configuration going over the limit, but if the load average calms down after a certain state, how can I ensure that there are healing services to make sure that PHP operations do not go down for hours.
If you have faced a predicament like this before, can you please confirm what guards have you applied on your end?