we have a few ESXi-hosts. They should be configured identically. However, one of our applications runs on two virtual servers that get migrated to different ESXi-hosts depending on the load on the physical host.
It seems that our virtual machines are "allergic" to one of the ESXi-hosts. Each time they are hosted on it, the CPU utilization for all processes goes up to aprox. 100%.
Once we move it to another ESXi-host, everything works fine.
All this is very new to me, i am usually on the Application side of things.
What/where do you suggest i should start troubleshooting?
Best Regards
Some CPU usage statistics include IO wait. That is: If the CPU works 20% and waits 80% of its time on IO, it shows as 100% CPU usage.
I've seen "high CPU utilization" like this before. Adding CPUs didn't help because the VM was mainly waiting for IO.
Check if the high utilization rate isn't due to IO wait. If it is, your host has some storage problems.
And another one: We have mostly HP DL380/385 servers. They have BIOS settings... I don't remember the wording right now but it boils down to low power consumption, balanced power consumption / performance and high performance. You should always use the latter when virtualizing. Maybe the "good" host is configured for high performance and the "bad" one isn't.