Using LEMP on popular cloud-service virtual server, with MyISAM tables.
The problem is that sometimes once in few hours simple mysql queries (select from where) executing several seconds (5-9) instead of milliseconds (0.002) like in other time.
MySQL slow.log has these records:
... Query_time: 5.369666 Lock_time: 0.000120 ...
error.log is empty and in other system logs have nothing special too
Such slow queries goes one by one to several databases and tables so its not looks like the table lock and lock_time value in slow.log is also insignificant.
I was enabling general log but there was nothing strange. Just regular queries so I cant find any correlation. I also didnt find out memory or cpu leak. Looks like sometimes something load MySQL a lot and it slowing down.
What can be the reason of such peaks of MySQL slowdown and how can find it out?