Remember when I said that if you had a 15 minute average greater than the number of the processors on your server? Well, today, in a server with just 2 CPUs, I had this situation:
Actually, at a point load average was even greater!!! And CPU percentage was smaller...
And yet, CPU percentage was averaging on 60%, and the server was not cloggy . What was going on?
In fact, I was executing a very heavy task, with lots of calculation and lots of IO operations on disks.
ANSWER
What I found was that there were an incredible lot of task (threads in this case) in queue, but most of them doing read/write operations on disk. For every operation on disk, each thread had to wait for some time; so, for this waiting time, in fact the CPU wasn't doing anything. Nonetheless, each thread is blocking the queue, and just waiting for their read operation to complete!!
So In fact there was "nothing" to worry about. How can I monitor this behaviour it's another matter.
