Davan

updated

13 years ago

D

Davan

28 Posts

0

5460

May 27th, 2013 15:00

Unusual MPIO Behavior

I have a PS 6000 series.

Connecting using ISCSI to a Windows Server 2008 R2 that I just build last week.

I have installed the HIT kit and set up MPIO with Round Robin.

I have created a volume.  For Access, I have limited connection by Initiator Name and by IP Address: 10.10.12.*

On the Windows server, I use the ISCSI Initiator to connect to the target four times, using the four NICs in the server:

10.10.12.208

10.10.12.209

10.10.12.210

10.10.12.211

I have saved these connections as Favorite Targets.

When I reboot the server, the connections automatically come back, however within a few minutes, these four connections are dropped and are replaced with:

10.10.12.208 and 

10.10.0.104

What the heck.  This last IP address is my data vlan, separate from the storage vlan.

I really can't explain why it keeps dropping the four connections I set up, and then substitutes with just those two connections, including one that isn't even on the storage vlan.

Any thoughts, comments, ideas or suggestions?

What's really weird, is this is the 2nd server in this infrastructure that I've built.  The first one is working perfectly as I expect it to - the connections to the proper storage vlan IPs stay the way I configured them.  But this 2nd server is behaving very unusually.

Thanks for any assistance you can give!

  • Origin3k

    6 Operator

    2421 Posts

    12396 Points

    407

    1

    Posted May 27th, 2013 15:00

    Iam not sure because we use HIT on VMware but i think there is an option to set the number of connection to a given volume. Take a look the the Remote Setup Wizard->Configure MPIO Settings.

    Regards,

    Joerg

  • Davan

    28 Posts

    407

    0

    Posted May 27th, 2013 15:00

    Hello, Joerg, Thanks for the reply. I didn't think about checking out the Remote Setup Wizard.

    In there I was able to EXCLUDE the 10.10.0.0/24 vlan, and add only the storage vlan 10.10.12.0/24

    This cleared up the issue with 10.10.0.104 being used as an initiator address.  Thanks!