PHP-FPM pool exhausted, or the site on a version it cannot run.
502 from a dead socket, a fatal after a plugin update, workers set higher than the RAM. We read the pool and the PHP log before raising pm.max_children.
The pool, the version, the log
Every worker busy, the site timing out. More children on a 2 GB VPS makes it swap. We look at the slow log and the RAM before adding workers.
A site moved to 8.3 that still calls a removed function. That account goes back to the version it ran. The others stay. A real upgrade is scheduled.
NGINX points at a socket the pool no longer creates, often after a panel rebuild. The error is the path, not a WordPress white screen.
We fix the pool. We do not rewrite the plugin.
Opcache, memory limit and the version are in scope. A fatal inside a paid plugin is reported, not patched. Theme work is not this job.
€55 an hour. Site down is emergency, subject to availability.
Minimum 1 hour. Emergency €95. A PHP major upgrade for every site on the server is agreed first, because one of them will break.
502 after a PHP change, or the pool maxed out?
Tell us the PHP version and the panel. No root password in the first message.