The server today had some difficult times - the load suddenly grows up to 5000%, but I did not find any local processes that cause extensive CPU load.
But I've found following in
vmstat 1 listing.
Here is a start of the lockup (see the last column):
And here is the end:
The manual refers to it as “Time stolen from a virtual machine“. More specifically:
It’s the time the hypervisor scheduled something else to run instead of something within your VM. This might be time for another VM, or for the Hypervisor host itself. If no time were stolen, this time would be used to run your CPU workload or your idle thread.
Here is another lockup at the same day:
At this time the server does not respond at all, SSH is almost dead, number of processes waiting for run time grow, requests from web to the Apache are not handled. And as you see after the end of lockup the number of interrupts per second and context switches is very high - it looks like the server tries to gain everything it missed during the lockup.
So the question - what could the problem be? This happened twice already - on this and last friday, almost at the same time. You could see it on a load graph:
i.stack.imgur.com/BuQ9k.png (sorry, does not have enough carma to post more than 2 links).