Powered by System Center
// Main banner ad on homepage
Archive for June, 2010
Find free disk space in Cluster Shared Volumes (CSV)
Jun 21st
Ahmed Bisht, Senior Program Manager from the Microsoft Clustering & High Availability team published a nice PowerShell script for exposing the free disk space in Cluster Shared Volumes.
Check out his blog at:
http://blogs.msdn.com/b/clustering/archive/2010/06/19/10027366.aspx
Removing HP Tools before or after P2V
Jun 21st
Here’s a handy tool from CtxAdmTools to have around when virtualizing HP servers to a virtualization platform. It cleans up most if not all of the HP tools, agents and programs.
http://ctxadmtools.musumeci.com.ar/HPPSPCleaner/HPPSPCleaner12.html
HP updates ProLiant Support Pack to v8.50
Jun 21st
HP has released a new version of its HP ProLiant Support Pack for Windows Server 2008 (both 32-bit and x64).
In general, a new ProLiant Support Pack contains important updates to storage and network drivers or firmware. Its main purpose this time however, is to support the newly announced HP ProLiant G7 series including its new iLO3 and fix a few minor problems.
If you encounter network instability in high traffic conditions, read until the end!
New is the support for the NC553i 10Gb 2-port FlexFabric Converged Network Adapter, which is capable of standard 1/10 Gbit Ethernet, iSCSI and Fibre Channel (FCoE) in combination with HP Virtual Connect Flex-10 Interconnects in HP Blade System enclosures.
Although there are no obvious improvements for Hyper-V hosts or clusters, I strongly advise to test this new version on less critical servers before implementing the PSP on Hyper-V production clusters.
The official upgrade requirement from HP: Optional
HP ProLiant Support Pack can be downloaded from:
http://bit.ly/9ZM1DI
After investigating if any of the network drivers had also been updated, I found several updates. Notably the HP NC_Series Broadcom 10GbE Multifunction Driver for Windows Server 2008 x64 and R2 x64, which had already been updated in March 2010, caught my eye. We use this one a lot in HP BladeSystems with HP Virtual Connect Flex-10 environments with Hyper-V R2 clusters.
It read: “This driver addresses an issue where the ethernet connection is lost or ping stops for some time during high traffic conditions.”
Official upgrade requirement from HP: at earliest convenience
In fact we have come across large scale implementations of Hyper-V R2 on this hardware with this particular issue. Hopefully this fixes this nagging and difficult to find issue.
Software and drivers for HP BL490c G6: http://bit.ly/a8WZS9
Hotfix required for Hyper-V R2 related to backup
Jun 17th
If you see prolonged startup time with Windows Server 2008 R2 servers with the Hyper-V role enabled, this can be caused by a large registry hive that increases after every backup.
“The issue occurs because of a large number of orphaned registry keys. The Volume Shadow Copy Service (VSS) snapshots create many registry keys. However, they are not deleted after the VSS snapshot operations are completed.”
The solution is hotfix KB982210: http://support.microsoft.com/kb/982210
SCVMM2008 R2 hotfix rollup
Jun 17th
A rollup is available for Microsoft System Center Virtual Machine Manager 2008 R2 which solves a couple of issues:
Issue 1: A high availability virtual machine with a GPT pass-thru disk is reported is reported as “Unsupported Cluster Configuration.”
Issue 2: When you manage VMware Virtual Center that has distributed switches, the Virtual Machine Manager service (Vmmservice.exe) crashes.
Issue 3: After Quick Storage Migration is completed, the Failover Cluster Management MMC snap-in reports that the VM still uses the old CSV volume.
Issue 4: If a VM is configured with 3 virtual processors, the SCVMM Admin Console crashes.
Issue 5: The settings for the virtual network are incorrect after a Refresh-VMHost cmdlet job is completed.
KB982522:
http://catalog.update.microsoft.com/v7/site/Search.aspx?q=982522
Additionally an issue with the SCVMM Administrator’s Console is solved:
Issue: If a VM is configured to have 3 virtual processors, the SCVMM Admin Console crashes
KB982522 replaces http://support.microsoft.com/kb/976246 which solved this problem:
“In Microsoft System Center Virtual Machine Manager 2008 R2, you remove a virtual hard disk (VHD) from a virtual machine. You expect that the .vhd file is detached from the virtual machine. However, the .vhd file is actually deleted on the Hyper-V server”
Update hotfix list Hyper-V R2
Jun 16th
I came across a new Hyper-V R2 hotfix on the list related to Intel Westmere processors (Nehalem EP/W5580).
“This issue occurs because Intel Westmere processors have a new Virtual Machine Control Block (VMCB) caching feature. This feature is added to provide a virtualization assistance to increase the hypervisor performance. Because this feature was introduced after Windows 2008 SP2 and Windows 2008 R2 were released, the hypervisor does not handle the VMCB caching correctly. Therefore, the system memory may be corrupted. This behavior can cause the stop error 0x0000001a.”
http://support.microsoft.com/kb/981791
The complete list of Hyper-V R2 updates can be found on the following Technet page:
http://technet.microsoft.com/en-us/library/ff394763(WS.10).aspx
Microsoft changes Hyper-V R2 cluster limits
Jun 16th
Until now Microsoft supported Hyper-V R2 clusters with a maximum of 64 VM’s per cluster node or 960 VM’s per 16 node cluster (taking into account a cluster reserve of 1).
With the upcoming SP1, these limits have now been increased to 1000 VM’s per cluster with any number of nodes, but with no more than 384 VM’s per cluster node. The maximum number of cluster nodes remains 16.
Where does the number of 384 come from? It is apparent that a dual socket quad-core totals 8 logical processors. If we take the 8 virtual processors to 1 logical processor ratio which is formally the maximum supported by Microsoft, it would take a quad-socket 12-core processor to get to 384 (4 x 12 x 8).
Please note that very high VM densities of VM’s in a Hyper-V R2 cluster requires a very high number of IO capability in the SAN. This is much more likely to be the limiting factor.
|
Number of Nodes in Cluster |
Average |
Max # VMs in Cluster |
|
2 Nodes (1 active + 1 failover) |
384 |
384 |
|
3 Nodes (2 active + 1 failover) |
384 |
768 |
|
4 Nodes (3 active + 1 failover) |
333 |
1000 |
|
5 Nodes (4 active + 1 failover) |
250 |
1000 |
|
6 Nodes (5 active + 1 failover) |
200 |
1000 |
|
7 Nodes (6 active + 1 failover) |
166 |
1000 |
|
8 Nodes (7 active + 1 failover) |
142 |
1000 |
|
9 Nodes (8 active + 1 failover) |
125 |
1000 |
|
10 Nodes (9 active + 1 failover) |
111 |
1000 |
|
11 Nodes (10 active + 1 failover) |
100 |
1000 |
|
12 Nodes (11 active + 1 failover) |
90 |
1000 |
|
13 Nodes (12 active + 1 failover) |
83 |
1000 |
|
14 Nodes (13 active + 1 failover) |
76 |
1000 |
|
15 Nodes (14 active + 1 failover) |
71 |
1000 |
|
16 Nodes (15 active + 1 failover) |
66 |
1000 |
Requirements and Limits for Virtual Machines and Hyper-V in Windows Server 2008 R2:
http://technet.microsoft.com/en-us/library/ee405267(WS.10).aspx
(Section Failover Clusters and Hyper-V R2)
Hyper-V R2 and Clustering Videos Technet 2010 North America
Jun 12th
Last week a number of great presentations were given at TechEd 2010 North America in New Orleans .
Here is a selection of sessions on Hyper-V R2 and Clustering that I consider to be my field. Some of them are presented by my Cluster MVP Lead Symon Perriman.
Disaster Recovery by Stretching Hyper-V Clusters Across Sites:
http://www.msteched.com/2010/NorthAmerica/VIR303
Failover Clustering Deployment Success:
http://www.msteched.com/2010/NorthAmerica/WSV313
Windows Server 2008 R2 and Microsoft SQL Server 2008 – Failover Clustering Implementations (by fellow cluster MVP Allan Hirt): http://www.msteched.com/2010/NorthAmerica/DAT407
Advanced Storage Infrastructure Best Practices to Enable Ultimate Hyper-V Scalability:
http://www.msteched.com/2010/NorthAmerica/VIR207
Hyper-V and Dynamic Memory in Depth (by Ben Armstrong):
http://www.msteched.com/2010/NorthAmerica/VIR304
Virtualization FAQ, Tips and Tricks (by Virtualization MVP Janssen Jones):
http://www.msteched.com/2010/NorthAmerica/VIR403
Networking and Windows and Windows Server 2008 R2 Hyper-V: Deployment Considerations:
http://www.msteched.com/2010/NorthAmerica/MGT206
Failover Clustering Pro Troubleshooting with Windows Server 2008 R2: http://www.msteched.com/2010/NorthAmerica/WSV314
HP Insight Control for Microsoft System Center (by HP Product Manager Doug de Werd): http://www.msteched.com/2010/NorthAmerica/MGT206
Windows Server 2008 R2 SP1: http://www.msteched.com/2010/NorthAmerica/WSV307
Hyper-V Backup: A look under the hood (Soumya Das Bhaumik)
http://www.msteched.com/2010/NorthAmerica/VIR322
Hyper-V R2 Component Architecture Poster
Jun 11th
I have seen several links to a great Hyper-V R2 component architecture poster:
http://www.microsoft.com/downloads/details.aspx?displaylang=en&FamilyID=5567b22a-8c47-4840-a88d-23146fd93151
But have you seen the Seadragon version?http://seadragon.com/view/12u8
Hyper-V R2 and Dynamic Memory in Depth
Jun 11th
If you are curious about how Dynamic Memory (DM) is implemented in Hyper-V R2 SP1, go and take a look at Ben Armstrong’s presentation at TechEd 2010 US.
http://www.msteched.com/2010/NorthAmerica/VIR304
Expect a public beta of Windows Server 2008 R2 SP1 towards the end of July. I can tell you it is a very well architected implementation of Dynamic Memory.
Ben explains how DM works under the covers and what you can expect from SP1.
Adding memory is implemented by a Synthetic Memory Driver (VSP/VSC Pair) which requires an update of the Hyper-V Integrations in the guest partition before it can use DM.
Supported guest operating systems:
-
Windows Server 2003 (32-bit & 64bit)
-
Windows Server 2008 (32-bit & 64bit)
-
Windows Server 2008 R2
-
Widows Vista and Windows 7 (32-bit & 64-bit only)
-
Enterprise and Ultimate Editions only
The new memory feature is clearly what it says: Dynamic Memory and NOT Memory Overcommit.
Memory Overcommit means Memory Oversubscription means Paging to the disk means Bad Performance
DM treats memory like Hyper-V treats CPU resources: as a dynamically schedulable resource.
So oversubscribing, NOT overcommitting.
New in SP1 too is the ability to protect the memory Root Reserve. By means of a registry key static memory can be set to reserve memory for the parent partition. This memory cannot be grabbed by guest partitions.
Ben also explains a new NUMA setting in the configuration of SP1 guest partitions.
Twitter
RSS
