Adding MirrorView disks on Windows 2003 Servers

Hi,

I have two ENC VNX5300 SANs

SAN_1, located at Primary Site
SAN_2, located at Recovery Site

Correspondingly, there are two File Share Servers (Windows Server 2003 R2 Std Edition)

File_1, located at Primary site
File_2, located at Recovery site.

Currently, a LUN on SAN_1 is presented to File_1 at the Primary site. Using MirrorView, the LUN on SAN_1 has been replicated to SAN_2 and the ReadOnly Mirror Image has been presented to File_2. On File_2, this Mirror Image disk is visible in Disk Manager but, is not formatted, partitioned, etc...

My issue is this, to verify a proof of concept for a disaster situation, the mirror was fractured and the ReadOnly Mirror Image was promoted.

Within disk Manager on File_2, before windows will allow you use the disk, it's insisting on formatting it. Obviously, this isn't an option as it will delete all the replicated from the Primary site.

I'm new to EMC Storage, having just migrated from IBM and, suspect this question should be asked on a Microsoft forum but, does anybody know how the Mirror Image disk, after being promoted, can be made available on a MS Windows server ?

Thanks

Responses(1)

DButch1

15 Posts

0

December 5th, 2012 15:00

I think you are seeing two problems:

MirrorView doesn't really present a read-only volume to the secondary host - it's basically unavailable for I/O while synchronized. (Also, Windows doesn't like read-only volumes.)
When you fracture the mirror and promote the secondary image, Windows has no way of knowing this has happened, and doesn't automatically dismiss the cached view of the volume and refresh it's LUN and volume information.

Note that Symmetrix SRDF treats things a bit differently - it does allow enough inquiry data on the R2 (secondary image) that Windows can at least ID the disk.

My memory of W2K3 interaction with replication is getting a bit fuzzy but try this - fracture the mirror and promote the secondary image. Reboot the secondary site host. If it properly picks up the LUN and volume you were hitting problem 2. Next step is to start with the volume synchronized, fracture it, and then try issuing a chkdsk /F on the server at the secondary site - IIRC the first thing chkdsk does is dismiss cache and do a dismount/remount of the volume on the LUN to make sure it has picked up the actual state of the physical media.

If this is going to be a production solution you might want to look at the AutoStart product - that helps create a more manageable distributed cluster and helps with the controlled failover and fail back, including leveraging some Windows features to deal with making sure the volume come up correctly. (MirrorView/Cluster Enabler performs a similar function but is only supported on Windows Server 2008 and above.)

View All

No Events found!