Unsolved
This post is more than 5 years old
69 Posts
0
9488
Windows 2008 Hyper-V running on EMC CX3-40f
Hi,
I have EMC CX3-40f external storage it is directly connected (without FC switches) with two servers PE-2950. Windows 2008 Hyper-V enable on both servers
I am using SQL on VM running on Hyper-V.
I created Raid, Storage group and presented the LUNs to these two servers.
I am able to see these LUNs in disk management, buts not all at the same time...i even restarted the machine but wasn't successful, it gets sort of hang some time...
If I remove these LUNs ..it works fine,
By the way i didn't initialized those disk on the physical servers.
Model of the storage: CX3-40f
Flare Version: 03.24.040.5.011.
Note: I want to upgrade the flare version also
Thanks & waiting for your response
Asim
Dev Mgr
9.3K Posts
0
April 29th, 2010 09:00
2 server can only share a LUN if they are clustered.
Also; due to the active/passive design of Clariion (I'll forgo on the ALUA option for this reply to your post), in a cluster, all servers need access to the same storage processor, or storage processors. In other words, you cannot have cluster server 1 connect only to SPA and cluster server 2 connect only to SPB.
So, my questions are:
- are your 2 servers clustered?
- is each server cabled to both storage processors?
- are you using the latest EMC certified HBA driver?
- which version of Powerpath did you install?
- are you running Windows 2008 with Hyper-V, or Windows 2008 R2 with Hyper-V
shkasim
69 Posts
0
May 3rd, 2010 05:00
- are your 2 servers clustered?
( Yes.. Active/Passive )
- is each server cabled to both storage processors?
( Yes..Direct attached without Fabric Switch )
- are you using the latest EMC certified HBA driver?
( Base driver, HBAnyware utility 4.1a35 )
- which version of Powerpath did you install?
( 5.3 with SP1 )
- are you running Windows 2008 with Hyper-V, or Windows 2008 R2 with Hyper-V
( Windows 2008 R2 with Hyper-V)
Note: One more thing I update the flare code ver: 03.26.040.5.031
Thanks for your answer till now still my problem is not yet fixed I am waiting for your response.
Bye
Asim
Dev Mgr
9.3K Posts
0
May 3rd, 2010 07:00
In Windows clustering, by design, a cluster shared disk can only ever be seen by 1 server at a time. This is why Windows clustering has always been active/passive, and active/active was only possible if you're running multiple applications or instances of applications (e.g. Exchange and SQL or 2 instances (databases) of SQL; each would be owned by a different node and they could fail over to the other if needed). The node that owns an application own the disk that the application has it's data on, so each application (database/mailboxes) has to be on it's own cluster disk(s).
Windows 2008 R2 added a new feature for Hyper-V called Cluster Shared Volumes (CSV). CSV is only usable by Hyper-V, but it allows all nodes in the cluster to actively access the same disk at the same time.
Some changes are important for CSV to work; in 'regular' clustering, for the heartbeat, usually you would disable everything except TCP/IP (or IP4 in Windows 2008). When using CSV, on the heartbeat you need to enable IP4, file and printer sharing and the Windows client.
Once this has been done, in cluster management, right click the cluster and select to enable CSV. Now you'll see an entry for CSV. Right click this and select to add a disk. You add the disk(s) that you want to use to store the VMs. After that you'll notice that these disks no longer have drive letters. They are now mapped to c:\ClusterStorage\Volumex. This is where you now store your VMs and both servers can see the 'disks'.