Hi.
I have pretty standard installation of FreePBX, based on the image distro, running on a virtual server with 1 vCPU and 2 GB RAM. There are 11 extensions, normally around 20 devices connected in total - so not too much going on. The highest number of simultaneous calls usually occurs when the conference is in use, then it’s around 4-5.
Since I started monitoring the server with the TIG stack, I can observe a weird pattern in the CPU loads (1 min, 5min, 15min). Every 6-7 hours, there is a “peak” and the load averages go up to between 1.4 (15min) and 3.5 (1min), as in the graph below (only showing 15 min average, and the number of CPUs at 1. The vertical lines are the alerts).
I have tried to find out what is going on that causes the loads to increase. I have not found anything. There seems to be no correlation with:
- number of calls
- number of connected devices
- network traffic
- number of processes
- memory usage
- crontab entries
If I try to manually monitor with top or htop during one of the phases, I can see that the total CPU usage percentage jumps around - however, I can not see that it is any one process causing this. I even checked with our virtual server provider if I could blame someone using the same vCPU, but apparently that is not the case either.
Does anybody have any ideas of what could be causing this? I’m out of ideas on what to check for, but just adding another CPU shouldn’t really be the answer!?