A bit more information is needed. How many VMs? How much RAM do you have configured for them?
Do you use ZFS?
The screenshot shows part of the node summary and as you can see, KSM sharing is munching 27.5GB of RAM together that is the same between some of the VMs.
From what I can see, the node is over provisioned regarding RAM. If something would cause to use more RAM (either a ballooned VM or KSM sharing cannot work as efficiently anymore) the node will run out of RAM and the OOM killer will "randomly" kill one of the VMs to get some RAM back.
As a general advice, either add more RAM or reduce the load on the node.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.