1

Once in a while (maybe once a day?) I've got a load average of 10. That lasts for like 30 min (where I can't reach any of websites on this server) and then it's fine again.
I just happened to be via SSH on the server and did a htop.
It's mostly php-fpm: pool www and /user/sbin/mysqld processes.

I've got Debian: Linux r3cwzg4q 2.6.32-042stab104.1
and up to date versions of PHP, Nginx, Mysql and all web apps. In total running 17 webs with very little traffic (0 - 200 visitors a day - mostly in the 0 - 50 area, only one web with 200).
It's a VPS, 5GB memory, 4 cores - normal load is around 0.3.

Is there a way to debug which web is causing this load?
Or if it doesn't come from a website, where would it come from?
(cron backup jobs only at night, not right now)

Improve this question
4
  • Other customers, maybe? It is an OpenVZ container, after all.
    – Michael Hampton
    Jan 9, 2016 at 16:08
  • If it comes from other customers on the server, my VPS would still show all load and CPU usage from every other VPS? Still, I'd be more comfortable knowing more (currently more like guess work :-S)
    – Chris
    Jan 9, 2016 at 16:12
  • Can you post logs from both products for the times in question? Agree that your VPS is isolated and shouldn't show other customers CPU usage. Seems like a heck of a big server for that amount of traffic.
    – Tim
    Jan 9, 2016 at 19:26
  • Start actively monitoring. Monit for example can warn you when the load reached a certain threshold. That way you can better establish patterns. Monit can even restart Nginx and/or PHP-FPM for you to remedy the situation.
    – JayMcTee
    Jan 11, 2016 at 19:56

0

Reset to default

You must log in to answer this question.

Browse other questions tagged .