Sharing a disk between virutal and physical Windows Servers

[chh_sys]

Hi all,

we have a small virtualized environment with a file server

for some 600GB of data in a (physical) RDM on a an HP MSA1000 array.

To improve backup time we would like tp present this same disk to our physical backup server.

When we try this with our test system we encounter several problems:

1. The data the backup server sees is not up to date. On one partition it sees data that is some days old on the other it does not see anything

2. From some random time on we get a massive flood of NTFS errors on this drive.

Now my questions:

- Is this sharing of a drive possible for normal data or just for a cluster quorum disk?

- What can be done to avoid these problems?

Any ideas how to achieve my goal?

Thanks

Christian

[rossb2b]

If your RDM is in virtual compatibility mode you can use VCB (vmware consolidated backup) to accomplish this goal. If it is in physical compatibility mode you may need to use your SANs native tools to make a snapshot or a clone.

[indrajeets]

You can always add the RDM and keeping the LSI LOgic controller bus sharing to physical would help you greatly !

[dmorgan]

Do you have VCB licensed? With this, you need to present the LUN on the SAN to a proxy server, usually your backup server. We use Fibre HBA's, and our tape backup library is Fibre attached as well, so it was easy to connect the Backup server to the LUN on the san through the Fibre switch. You install VCB on that backup/proxy machine, and it allows you to mount windows machines on a file or fullvm level, and all other OS's at a fullVM level. So long as you on in a Windows environment, you can do a file level mount, and VCB will handle the snashot, and mount the file systems as folders on your proxy machine. You can then back up these files this way, and open files are managed due to the snap shot. This assumes you are running a VMFS file system.

If you found this or any other post helpful please consider the use of the Helpfull/Correct buttons to award points

[Texiwill]

Hello,

In order for the 'shared disk' to be up-to-date between the VM and the physical server you need to be running a clustered file system and most likely MSCS as well. As for getting a backup, the VCB approach discussed works quite well.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education. As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization

[cmanucy]

I'll warn you right now that when we introduced VCB to our MSA1500, we started having all kinds of problems - controllers locking up, lost drives, etc...

So proceed carefully, and if you can, limit I/O on the LUN to avoid SCSI reservation conflicts. Hopefully you won't have the issues we did with ours.

----

Carter Manucy

[polysulfide]

There are some other posts with good feedback on backing up.

NTFS is not a clustered filesystem. You need one host to release the disk and then the other host to refresh the mount before you'll have accurate data. Meantime the orginal host isn't able to use the volume. MSCS facilitates this process.

You can always run NTBACKUP in a variety of creative ways too.

If it was useful, give me credit

http://communities.vmware.com/blogs/polysulfide

VI From Concept to Implementation

[jhanekom]

With the number of times this type of question pops up in the forums (usually after people have already tried it), I think we should petition for a feature change that prevents you from powering on a VM with a shared disk unless you've explicitly disabled a warning that file system corruption might result if the file system isn't clustered. Of the "only do this if you understand the implications thoroughly" variety.

[cmanucy]

No offense to you or anyone else, but I do believe this, just as many other topics around here, qualify for "RTFM".

Personally, I wouldn't want 50 pop-ups warning me I'm about to do something if I know what I'm doing. If I don't know what I'm doing, that's my own damn fault.

----

Carter Manucy

[dmorgan]

I learned my leason before I started working with VMWare, when we first got our SAN. I figured I could just mount up an NTFS LUN from our tape backup device and back up the files that way. Three days later, after the data in that partiular database was corrupted and/or lost, and after rebuilding that entire file system and re-importing and updating one very large oracle database, I realized the hard way that NTFS is not aware that more than one computer could possibly be accessing it at the same time. VMFS, however, is multi-system aware. Thus, although many ESX servers may be attaching to the same LUN, only one has control over certain files at any given time. Plus, with the snapshot feature, you can mount up part of the VM as read-only, while the system is up and running, and access those files from additional machines.

If you found this or any other post helpful please consider the use of the Helpfull/Correct buttons to award points

[polysulfide]

ESX and VC require a multitude of cross-discipline compenetcies. I would rather people be forced to acknowledge this once during install or on a support agreement that to have to do a competency check at every option.

If it was useful, give me credit

http://communities.vmware.com/blogs/polysulfide

VI From Concept to Implementation