Windows 2000 Server with Fibre Array Problem

[James Smith_12]

I have 2 Proliant 1850R Servers Running Windows 2000 Server... I have 2 Storage Works Fibre Channel Array controlers in each Server with redundant connections running out to redundant Fibre Switches, and then on to the Fibre Array...

Here is my problem... Both Servers see the array as E Drive...but only one see's it correctly... One can read and write fine but the other only see's the partition correctly after a reboot.. They don't share it. they seem to have contention over the partion.. am I missing some thing to get these servers to coexist with this SAN?

[Eugeny Brychkov]

Disk array has LUN(s). These LUNs are presented to OS as disks (in disk administrator). If you want to share LUN between 2 servers then you need MSCS (clustering software) and define this LUN (disk) as a cluster disk.
If servers do not share one LUN but one server can access one LUN fine but another can not access another LUN (correctly as you said) then I suspect something is wrong in LUN's configuration within disk array OR in host configuration (driver, driver settings).
BTW, what's this disk array? What do you mean server can not see partition correctly until you'll reboot system? Any more techincal details?
Eugeny

[James Smith_12]

Here is all the info I can think of...

1:If I log into server A that does not have a problem, I can transfer files in and out fine... but If I transfer a test file in to the array Server B does not see it... but If I reboot Server B file access seems fine with server B but it is now server A that has the problem.

2:Hardware is fine because If I connect the array to the 2 Compaq 1600 Servers I have running Win2003 Enterprise server I can create a cluster fine.....

3: If I look at Disk manager for Win2003 enterprise Server, the Disk Manager See's the Array twice but one is greyed out but In disk manager in Win200 Server the array only shows up once. It's almost as if the OS does not See the redundant path to the Fibre array.

I enclosed A picture to help you see what I mean

4:If I go to Server a and delete a test file and then go to Server B the file still shows up but it's inaccesable.. If I reboot server B the file no longer shows up.

5:I can get one server or the other to see and use the fibre array correctly but not both at the same time.

[Eugeny Brychkov]

James,
looks like it's clustering question. I've heard that if one server in MSCS cluster grabbed cluster disk then another server(s) have no access to it. So from what you're writing it could be correct behavior you see... just use shared cluster disk only on one node which runs package. If this server fails (or you manually reboot it) then another server takes over.
I hope someone from MSCS experts here will tell you more
Eugeny

[david5337_1]

You need do this:
The first,if you used redundant path to connect the storage works,you need to use the secure path software to manage the paths;
The second,you need to set the LUN masking.

[Christian Schwartz-Søre]

In MSCS the cluster disks are owned by a node. That node (A) has full access, the other node (B) in the cluster will not be able to see nor access the cluster disks, before a failover occurs.
Once the failover occurs, the (B) node will have access to the cluster disks, and they will be present in the disk administrator.

This is normal behaviour.

Also with dual links you need Secure Path on the hosts to present the LUN's correctly, and to provide failover.

Regards
christian