Unsolved
This post is more than 5 years old
1 Rookie
•
1 Message
3
203618
December 4th, 2014 06:00
Dell Power Edge R320 - Windows Server 2012 R2 - Hyper V - High CPU Utilization - Performance Per Watt
Dell PE R320
Windows Server 2012 R2
Hyper-V Role w/ 4 VM's
IDRAC 7 / BIOS Version 2.3.3
1) Intel(R) Xeon(R) CPU E5-2420 v2 @ 2.20 GHz E5 220 MHz
Issue/Symptoms: System online in production approximately a month. No issues during that time. One morning we received SCOM alerts on high CPU. Upon investigation, the Hyper-V Host CPU was averaging 25% with spikes to 50%, the VM's CPU average 80% spiking to 100%. Memory is not an issue on either Host or VM's. No system changes were identified. System performance severely degraded, users report slow access to network shares, users directories.
Troubleshooting: Rebooted (several times), upgrading NIC Firmware/Drivers, uninstalling McAfee, Backup Exec and the SCOM Monitoring Agent with no results. We were leaning towards it being a VM issue on the host so we opened a case with Microsoft. Throughout 10+ hours we supplied traces using Perfmon, Process Explorer and XPerf which were all inconclusive. It was grueling getting the traces and reports to complete because of the degraded system performance.
Resolution: Logging into the IDRAC, booting to the BIOS, select System Profile, change the profile from Performance per Watt to Performance.
Immediate results... Hyper V Host CPU dropped to 0% - 4% average. It was like a new system.
The posts in this forum by user "LSLC" assisted immensely towards a resolution.
DELL-Chris H
7 Practitioner
•
9.7K Posts
•
48K Points
1
December 4th, 2014 09:00
Marthac,
Thank you for taking the time to share the issue you experienced, as well as the solution, with the forum. As LSLC's post had assisted you in finding the resolution, I am certain this will also help others.
Thank you again for your contribution.
dml6122
1 Message
1
December 10th, 2014 13:00
Great notes. Encountered the same issue with an almost exact timeline. Resolution posted by Marthac worked on our R320 experiencing exact symptoms.
Thanks
Azuziel77
1 Message
1
January 27th, 2015 07:00
This post just saved me. I'm actually running a PowerEdge T410 at a client. Server 2012 R2, hyper v role, running 1 x 2012 R2 exchange 2013 vm - simple setup. About a month ago, this same situation occured to client. Host suddenly using 40% cpu, and vm's, even recovered/cloned VM's were all running 100% CPU. Exchange Services wouldnt even start, not because they couldnt, but because system was so slow they literally wouldnt get to the point the system starts them after boot. EVen starting manually, and waiting hours wouldnt make a difference. Recovered a vm from our veeam backups, and it behavved the same way, 100% CPU, no exchange, mass mail outage. Anything I killed, and another process would eat the CPU. Happened again yesterday, about a month after the first incident, which seemed to resolve after reinstalling the CPU Drivers. tried reinstallign drivers yesterday, no joy. Came across this article, and noitced CPU in my T410 is same. Figured I'd give it a shot. After swapping performance per kw to performance in the profile area of the bios as instructed, server performance is now incredibly fast. This saved me hours and hours of troubleshooting, and a Dell ticket. It isn't just on R720, T410, R320, it seems to be e5 series processors and dell bios/lifecycle . Hope this helps anyone else that comes across this. As an aside, I'll be turning this setting off on other servers to see if the performance increase is anywhere near as dramatic. the t410 is like a brand new server again. Never ran so well.
pcmeiners
6 Operator
•
1.8K Posts
1
January 27th, 2015 09:00
Martha
Interesting issue, thanks for posting.
asdfad
1 Message
0
April 10th, 2015 16:00
Just for the next poor soul that comes across this thread. We tried the above fix and it helped very little. In the end we powered down the server, removed power cords, and held the power button for 10 seconds. Powering up the server next time resulted in a whole new animal and everything is zippy.
Casey
Ken Rappold
1 Message
1
April 16th, 2015 13:00
Making the change in my environment did more than resolve high CPU usage. We were also seeing event 533 in the Directory Services log. Making this change helped diminish those log entries. If I find another setting that removes 533 log entries altogether, I'll report back.
mono1940
1 Message
0
August 24th, 2015 12:00
same situation here.... I needed a reboot as ASFAD ( casey) noted, we needed to add the reboot steps to fix the situation, and the previous steps.
Thanks,
Gus
Troy D. Hector
1 Message
0
January 13th, 2019 22:00
Can you send me a step by step on how to do this?