Best Practices for Migrating Hyper-V Storage (SAN LUNs) for Microsoft Exchange Server 2010 | Recommendations for Migrating SAN LUNs for Hyper-V Server, where Exchange Server 2010 VMs are using Pass-Through Disk for Mailbox Databases

Author: Zahir Hussain Shah, MVP Exchange Server

Best Practices for Migrating Hyper-V Storage and SAN LUNs for Microsoft Exchange Server 2010 VMs and Hyper-V Pass-Through Disks for Microsoft Exchange Server 2010 Mailbox Servers Mailbox Databases

This article is in correlation with my earlier article for Expected issues after upgrading EMC Power-Path software for Hyper-V Server to access SAN Storage for hosting Microsoft Exchange Server 2010 VMs, for VHDs and Pass-Through Disks, in which I documented some of the problems and issues, we faced while migrating the SAN Fabric for Hyper-V Server, and with this article I will continue and will conclude this series of blog post with some of the best practices, which I learned for migrating our SAN Fabric from EMC CLARiiON to EMC VNX, where we moved the SAN storage for couple of Hyper-V Servers, including few of them, where we are running Microsoft Exchange Server 2010 Hyper-V Virtual Machines.

Problems and issues assicated with SAN Migration for Hyper- Server VMs:
Either you have simply assigned a storage LUN to a Hyper-V Server, where you are keeping the VHDs for the VMs, or you are using the SAN LUNs to be taken inside the Hyper-V Virtual Machine, by making them offline on the Host (Hyper-V) and add Physical Disk into a SCSI Controller, to have the physical SAN LUN assigned to the VM, when you perform any kind of upgrade activity on the SAN, I mean either you upgrade the EMC or any other software on the Server (Hyper-V), which provides multipath functionality, like EMC PowerPath, where EMC PowerPath provides the Server to see the LUN as a single DISK on the Server, instead of seeing it as a multiple disk, because usually the Servers are zoned from the SAN for multiple SAN Switches using multiple cables (in case of FC), so without using such MultiPathing software, OS does not know that whether its a same assignment for me or there are actually two different disks.

So as being said, when you perform the EMC PowerPath upgrade for firmware, either you perform in-place upgrade or uninstall and reinstall, or while migrating the SAN System, lets see from EMC CLARiiON (OLD VERSION) to the EMC latest VNX SAN, if you perform SAN COPY or use EMC OpenMigrator, when you will provide the new LUNs with the copied data, then Microsoft Windows Server recognizes these LUNs as new and different LUNs, and what Windows does here, it changes the UUID / GUIDs of the DISK, and which turns as a disaster for the Hyper-V, because when after migrating the SAN LUNs, and your storage administrator provides the new LUNs to your server, and you go to either turn on the VM in Hyper-V or you go to the Hyper-V Virtual Machine Properties, you will see that the SCSI Controller Hard Disk for Physical Disk (Pass-Through) will show you as NO DISK, and the order of the Pass-Through Disk will forgotten by the VM.

Solution / Recommendations for Migrating SAN LUNs for Hyper-V Server, where Exchange Server 2010 VMs are using Pass-Through Disk for Mailbox Databases:

Recommendation I: Note down the order and the disk names of the SAN LUNs on the Hyper-V Host Server, because it will help you a lot, once you will finish the SAN migration and reconfigure the Pass-Through Disks in the Hyper-V Virtual Machine for the same old order.

Example:

Disk Name on Hyper-V Server Disk Capacity (Size)
1 DISK 3 245 GB
2 DISK 4 120 GB
3 DISK 5 50 GB
4 DISK 6 245 GB
5 DISK 7 130 GB
6 DISK 8 100 GB
7 DISK 9 100 GB
8 DISK 10 200 GB

Recommendation II: It is also advisable to either take the snapshot or note down the Hyper-V VM Disk Configuration, as follows:

Example:

SCSI Controllers (Total SCISI Controller 4)

DISK CAPACITY (SIZE)

DRIVE LETTER INSIDE VM

DISK VOLUME NAME INSIDE VM

SCSI Controller 1 (MEMBER DISK):
DISK 3 245 GB O:\ 1_DB
DISK 6 245 GB P:\ 2_DB
SCSI Controller 2 (MEMBER DISK):
DISK 4 120 GB Q:\ REMOTE_SITES_DB
DISK 7 130 GB R:\ MANAGERS_DB
SCSI Controller 3 (MEMBER DISK):
DISK 5 50 GB S:\ MGMT_DB
DISK 8 100 GB T:\ IT_DB
SCSI Controller 1 (MEMBER DISK):
DISK 9 100 GB V:\ USERS_DB
DISK 10 200 GB Y:\ REOCVERY_DB

Tip: Before your shutdown the Hyper-V Server for letting your SAN Admin to cut-over to the new SAN, make sure that you DISABLED THE HYPER-V Windows Services on the Hyper-V Server, which will prevent the Hyper-V Node to boot-up, while the disk are in offline mode.So lets say you have noted down all the VM Pass-Through Disk Configuration and Hyper-V Disk LUNs disk associations, now your SAN Admin will ask you to shutdown the Hyper-V Server, so I can swtich over to the new SAN.

And once the SAN Admin (I call them penny-pinching people) done with his work, and will ask you to power-up the server, make sure that you ENABLED and START the Hyper-V Services, if you stopped and disabled them before shutting down the Hyper-V Server, okay, so now you booted up the Server, and will go to the Disk Management, you will see all those SAN LUNs, where you were keeping the VMs VHDs, are offline, and all others DISKs, which supposed to be offline on the Hyper-V Host, because we wanted them to stay offline, because we have taken them inside the VM as Pass-Through Disk, they will also be offline once the server boots-up, ok so whats the problem, problem happens, when your Hyper-V Server boots-up with the boot-up of the Server, it will try to start all the VMs on the Server, and since the Hard Disk on which the VHDs are stored, is still offline, so the Hyper-V will not show you that machine in the available VM list on the Hyper-V Node (dont freak out here :$, that where the hell your VM gone J), bring up your VHD SAN LUN disk from the Windows Disk Management, and refresh the Hyper-V Console, and you will see the VM back (By the way, it happened with me that to some how when the Hyper-V started and the disk on which I was having VHD, was offline, it deleted the VH Configuration Files).

So as best practice, before you started the Hyper-V Services or Hyper-V VMs, you ensured that all the Physical Disk on the Hyper-V Node are in their perfect fashion, that all the disk on which you are keeping VHDs are online and all others which are configured as Pass-Through for VMs are offline on the Hyper-V Node, here you can verify the Disk Numbers and Disk Capacity, with the tables mentioned above.

Recommendation III: Once we verified all the disk and their orders, now its time to visit the Hyper-V VM Configuration Settings, and as explained above that you will see that the Pass-Through Physical Disk Configuration inside the VM will be configured, means SCSI Controllers would be there, along with the DISK added into them, but the Disk Configuration means, which VM Disk is mapped to the which Physical Disk.

Since we do have everything on paper or in snapshot (image mspaint you rock), so we can just go to the Exchange Server Mailbox VM, and we can select each Disk and select the corresponding Physical Disk.

Once the all the SCSI Disk are corrected, you can turn on the Exchange Server Mailbox Node, and you will see it will turn on without any problem.

Remember, once you will do the same thing, all you have to do is to go to the Disk Management in the Exchange Server Mailbox Node (or any server), and bring the Disk Online, and you will see all the Microsoft Exchange Server 2010 Mailbox Database Disk will be back, and mailbox replication can be resumed from the Microsoft Exchange Server 2010 Management Console or shell.

Cheers!

zahir_sig_rahat_pc_1.png

About these ads

About Zahir Hussain Shah

Systems Infrastructure Consultant, Systems Solution Architect, Senior Systems Engineer. Certifications: MCSE, MCTIP Entperprise Administrat

Posted on 04/01/2012, in Exchange 2010, Hyper-V, Server Migration and tagged , , , , . Bookmark the permalink. 2 Comments.

  1. I’m running into problems where the pass-thru disks for my Exchange mailbox server VM will not be offline and are missing from the VM config after a reboot of Hyper-V. These are direct attached disks on D2700 attached to a DL380 Gen 8.

  2. Try the following:

    1) First of all ensure that your host and guest machines are fully patched, and for check this you can download OS related Hyper-V and cluster hot-fixes from here: “http://zahirshahblog.com/2013/01/27/exchange-server-virtual-machine-slowness-exchange-server-performance-problem-hyper-v-slowness-for-exchange-server-virtual-machines/”

    2) Also check that there is no program and script running which is rescanning or bringing the passthrough disk online on your physical Hyper-V hosts. If the disks remain offline in the same order, no matters how many times you restart your Exchange mailbox guest machines, the disks should always be added in the VM and should come normally in the VM after restart.

    I believe it could be a bug in your Hyper-V servers, try to install the hotfixes, it should get resolved. I saw from the real-world standpoint Windows Server 2008 based Hyper-V release had lots of performance and VM availability related bugs in the product.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 312 other followers

%d bloggers like this: