I have a couple of nodes ( which are not mine ) running one openvz kernel version -2.6.32-042stab092.2

specifications : processor model name : E5-2620 0 @ 2.00GHz Number of Processors : 24 RAM : 48G number of VPSs hosted on each : 23

each vps is assigned 1000 unit of cpu and a cpu limit of 400 each vps is assigned 1G of memory

after some researching i have found that running on el6 kernel that means that each vps can take up to 1000/400 each cpu running intensive processes .. which means a total of 25% if the vps is running on maximum processing , am i right ?

now i face a problem with high load , some of the vbs are running forums with plugins enabled and intensive mysql access .

problem is whenever a VPS is causing a high load the whole node is also affected by it and the load average raise , which causes other VPSs problems .. slow them down

why is this happening ? apart from resource management inside the vps it self , how do i prevent one vps causing load to not slow down the whole node and raising it's load average ?

Thank you for your time


The load average signifies the amount of processes in a runnable or uninterruptable state. So even with the CPU time limited with --cpulimit, the running processes will still be counted towards the load average.

There are also many other reasons load could be high with little user CPU usage. Interrupts and disk wait are contributors to load. Interrupts are common with higher network loads and disk wait is when the CPU is waiting for the disks to respond.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.