Time for a quick-and-easy article about an interesting scenario we came across recently, after deploying a new service for a customer.
The service we deployed was x2go, which is a VNC/Terminal Services type experience for users running Linux. This particular experience was CentOS 7 with the XFCE Desktop environment, and the issue that we experienced was XOrg constantly utilising 100% CPU, as you can see from the following screenshot:
Now, as this was a new server build the pressure was on us to find a solution, and it was a strange one as the server was pretty standard, and the old x2go server was running on a similar spec (t2.large, as opposed to the t3a.large of the new one).
Then I had a thought – I had changed the instance type from a t3.large to a t3a.large, and some reading on the Internet pointed me to an IRQ/Hardware issue with Ethernet Adapters and/or Intel graphics (AWS Instances don’t have Intel graphics, so maybe Ethernet..?).
After a quick shutdown and type switch, the CPU Utilisation issue went away. This means that ultimately –
Using as t3a.x EC2 Instance seems to cause 100% CPU Utilisation on Xorg.