1

Recently I've had to confront to php-fpm issue with constantly growing of available active process. It was reaching max_children and then restarted.

The graph shows "active processes" value from pool status from three servers (round robin balance).

active processes

The configuration:

pm.max_children = 100
pm.start_servers = 20
pm.min_spare_servers = 5
pm.max_spare_servers = 35
pm.max_requests = 2000

At 01AM I changed configuration to:

pm.max_children = 500
pm.max_requests = 500

This pool serves about 4k req/min (for all servers). I must admit other pools are working normally with the same settings (biggest pool serves ~12k req/min).

What could be a reason for such weired behavior?

Improve this question

1 Answer 1

Reset to default
1

I think that you should enable "slowlog" and "request_slowlog_timeout" in your pool configuration. Seems like your backend stucks on some hard requests. Maybe because of long database requests or disk I/O

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .