2

I have an odd situation here, so I thought I'd see if anyone else has run into it. I have a Hyper-V host (2012) that had two VM's on it, one of which was an Exchange 2013 server. The other is a 2012 file server. The Exchange server has been pegged at 100% utilization, the file server is about 50%.

As a test I built a Windows 2012R2 server and being a fresh install doing exactly nothing it had CPU utilization of 10-15%.

I have been troubleshooting with MS thinking it was something on the Exchange server causing the issue. But today I moved it (using Hyper-V move while it was running) to another (similar hardware) Hyper-V host, and the utilization has dropped to 2%.

The hardware is roughly equivalent but not exactly the same.

  • Is the Exchange server active, or is it just for learning purposes? Are the mailboxes stored on the VM? – MDMoore313 Feb 13 '14 at 20:03
  • 1
    The server is one half of a DAG, and it's in the first phase of a general migration from Exchange 2010. The mail DB's were active on the other node of the DAG because this node was not able to keep up with the load due to the high CPU condition. The mail DB's are stored on the VM (separate vhdx from the OS). There are only 3 active users on the new mail system. – theunpaidBill Feb 13 '14 at 20:07
  • How is the VM provisioned? Did you follow the MS guidelines as per the Exchange Server 2013 hardware requirements for a virtualized Exchange 2013 server? - http://technet.microsoft.com/en-us/library/aa996719(v=exchg.150).aspx – joeqwerty Feb 13 '14 at 21:02
  • What processes are hogging the CPU? – RobbieCrash Feb 13 '14 at 21:40
  • Yes I followed the MS guides, note that the VM runs at 2% utilization on the other Hyper-V host, so I don't think it's a problem with the VM. – theunpaidBill Feb 13 '14 at 21:46
  • The Hyper-V host seems to be running at about 25-30% utilization, which is really quite high considering it's down to 1 VM which is only doing file serving. Unfortunately there doesn't seem to a particular service that is eating the cpu cycles, everything seems to be taking more than normal. Heck the Event log service has spiked up to 8-15%. The VM worker processes seem to be running at around 1% utilization, everything else is just normal base server processes. – theunpaidBill Feb 13 '14 at 21:50
  • As a test I moved a clean Windows2012R2 server, base load, no features enabled, just straight from the install from a Hyper-V host to the problem host. CPU utilization on the guest went from 0-1% to a steady 25-40%. The host utilization did not change appreciably. – theunpaidBill Feb 13 '14 at 22:14

2 Answers2

0

MS support was unable to isolate a cause for this condition;however, migrating all vm's off of the host, and then back fixed the utilization issue. I can only assume that it was something in the way the VM's were interfacing with the physical processors and the migration process eliminated it.

0

Check your bios. Make sure to set performance to maximum instead of power saving mode. Some Dell servers have that issue.

Loc
  • 1
  • That wouldn't cause the VM to have high CPU utilization, it would (could) prevent the VMs from using all the hardware resources that are available. – HopelessN00b Aug 20 '16 at 23:48
  • 2
    @HopelessN00b — Actually, it would (and does). You can read more about it [here](https://serverfault.com/q/593795/218519). – InteXX Jul 26 '17 at 17:51
  • @InteXX I stand corrected. Yet another argument against power saving mode on a server, I suppose. – HopelessN00b Jul 26 '17 at 18:10
  • @HopelessN00b — It sure had me questioning my sanity. We tried *everything*. And in the end it was such a simple solution—exasperating. – InteXX Jul 26 '17 at 18:17