Win2003 host running MSSQL is having some weird CPU issues, help.

[Elvar]

Hi,

We have been running a few MSSQL servers in our VMware enviroment. Everything went well when we were running 3.0 and 3.01 but after upgrading to 3.5 things have started to become crazy.

The hosts are what seems to be having huge cpu problems. Sometimes the sql process gets stuck in 100% for few hours and the host is usualy in 50%+ for days. So far we have found nothing wrong with the SQL queries and this is happening even if there is no activity on the server.

So before we start to pull our servers from VMware are there any known issues or something we can do to fix this?

Hardware is

IBM Bladecenter, LS21 blades, 10-16gb memory each

IBM DS6800 SAN

[gary1012]

Does the server and SQL respond during these spikes or is it nearly unusable? The reason I ask is because if it responds and is usable, perfmon might be giving you erroneous numbers. Perfmon was never designed to deal with way time is presented to a VM. If you're using perfmon, you may want to try vmperfmon. You can get it here: http://www.run-virtual.com/?page_id=156

[Elvar]

It depends if the cpu is at 100% or less. We often get timeouts. Only using task manager and viewing performance in VC.

We did create a new vm dedicated just for one database, the database is only 1gb or so and the machine has 3gb of memory. Still this happens.

I'm currently creating a raid 10 array and I'm going to storage motion the machine onto that array, just to rule out raid5 as the problem (could very well be)

edit,

I ran vmperfmon and the red line comes up as red at the top and this is from the cmd windows

cpuReservationMHz: 0

cpuLimitMHz: 4294967295

cpuShares: 2000

cpuUsedMs: 12751607

hostMHz: 2200

memReservationMB: 0

memLimitMB: 3072

memShares: 4294967293

memMappedMB: 3071

memActiveMB: 276

memOverheadMB: 191

memBalloonedMB: 0

memSwappedMB: 0

memSharedMB: 352

memSharedSavedMB: 278

memUsedMB: 2793

elapsedMs: 22426461

resourcePoolPath: 'host/user/vm.2284'

Current Cpu Usage: 125

[kjb007 · Accepted solution]

You may be hitting an issue with the vm overhead growth limit. Check this and see if it will help your vm: http://communities.vmware.com/thread/119697 & http://www.vmwarewolf.com/vmotion-in-35-drs-enabled-cluster-causes-guest-cpu-to-rise-dramatically/

You can adjust a setting, and it may fix your scenario.

-KjB

[Elvar]

That seems to be the case with us, thanks.:D