WAMSINC
asked on
how to track down high CPU on a 2008 server
hello experts we have a 2008 server that is hosting 3 virtual machines suddenly start having increased CPU at 25-50%. This servers virtual machines are very slow to respond. Logging in was slow and each click took several seconds. All the hard drives on the dell server are green. Task manager shows a few items with about 10%, nothing really high. Whats the best way to get to the bottom of an issue like this?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
I checked for snapshots and nothing running. I had an old snapshot on one machine that I deleted and no effect. I turned off all VM's and rebooted a couple times and still very slow on each click. Im trying to get process monitor on there now.
It is very important that any VM's that you are using to run servers have sufficient memory. I had a need to run a pair of WebSphere servers, An AD Node, an SQL server and a desk top all on VM's in the same host PC. I allocated 4 Gb to each WebsSphere VM, 4Gb to the SQL 2008 server and 1 GB to the AD server---a total of 14 GB just for the VM's. I had to bump the memory on my host laptop to 24 GB: before I bumped the memory the VM's ran so slowly that it appeared that they had crashed.
For most modern CPU's you should have plenty of CPU power (I have 8 cores) and with VM's it is almost always a memory bottleneck. The real problem with insufficient memory is that the VM's end up thrashing in swap space.
Do make sure the host CPU is not bogged down but that is rarely the case unless there is an infinite loop someplace or you are running something really pegs the CPU like compiling a large Visual Studio project.
For most modern CPU's you should have plenty of CPU power (I have 8 cores) and with VM's it is almost always a memory bottleneck. The real problem with insufficient memory is that the VM's end up thrashing in swap space.
Do make sure the host CPU is not bogged down but that is rarely the case unless there is an infinite loop someplace or you are running something really pegs the CPU like compiling a large Visual Studio project.
ASKER
this is a dell r520 with 65GB of RAM and 12 CPUs Running 2008 R2.
ASKER
Task man shows CPU at 25-40% after shutting down all VM's
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
so mysteriously the cpu spikes are gone and its staying around 9 % now. It was acting like a failed disk but all the lights were green. Now I can move around the OS faster and clicks dont take forever. Im concerned this could happen again
Look at scheduled tasks. Depending on what your VMs do, data intensive application, exchange, SQL, data aggregates and begins to consume more and more resources before the first access is even seen.
Underestimating the resource requirements might explain I.e. Obe VM experienced higher than expected demand.
Underestimating the resource requirements might explain I.e. Obe VM experienced higher than expected demand.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
I had all the VMs off and it was still spiking CPU. I will check out PAL. Im looking at AV and windows updates. Task Scheduler was a good idea Im looking at the history on that as well.
Specs of host memory, CPU drives.