Solved!
Go to SolutionMD3000i virtual disk not on preferred path
Our MD3000i keeps saying the array is not optimal and that a virtual disk is not on preferred path.
We have two RAID sets on it with one VD each. The first says its preferred path is Controller 0, the second says its preferred path is controller 1. But controller 0 likes to take over both.
Using MD storage manager, I can change VD2 to be owned by controller 1. But sometime in the next several hours, it switches back, and the MD3000i starts blinking amber again.
If I reboot the server that is attached to it, the issue usually starts without having to wait a few hours.
I've changed the Ethernet cables out, and it didn't help. Incidentally, this all started when we changed the hosts that the MD3000i is attached to. It used to be attached to two Dell servers, now it is only attached to one.
DELL-Kenny K
685 Posts
1947
0
Posted October 22nd, 2013 10:00
Wardog25,
The best thing that we can do would be to have you pull a support bundle from inside the MDSM software. I will be sending you a direct email so when you get the report you can email it to me directly and I will review it and see what is causing the VD to not stay on the preferred path.
DELL-Kenny K
685 Posts
1947
0
Posted October 31st, 2013 14:00
I have not. Can you please try and resend?
DELL-Kenny K
685 Posts
1947
0
Posted October 31st, 2013 18:00
The first thing that I would look into would be to make sure that you can ping all of iSCSI ports from the host. That way we can make sure that the port is accessible and able to send and receive data. Also what OS is running on the host?
You were right, one wasn't ping-able.
I didn't set this system up, so I'm kindof learning as I go. Currently our attached server has 2 NICs that run to the MD3000i. One goes to Controller Module 0, Port 0. The other goes to Module 1, Port 0.
Strangely enough, the setup had M0P0 and M1P0 on the same subnet, while M0P1 and M1P1 where on a different subnet. I would have thought if you were going to do a separate subnet, you'd separate the Controllers. Is that normal?
Anyway, my 2nd NIC that goes to Module 1, Port 0, had an IP for the second subnet, not the first. So I changed that. I can ping it now. So I changed ownership back to Module 1 and we'll see if it sticks.
Ok, it still is giving the same problem.
I changed the IPs so that each controller is on a different subnet. I verified that the 2 NICs are on the correct subnet for their controller and made sure the iSCSI initiator was pointing to the right target IPs.
I can ping those target IPs just fine now and still when I set the 2nd virtual disk to be owned by controller 2, it seems to correct the problem for 5-10 minutes, then it defaults back again and the array changes to non-optimal.
Sigh.
Yes, it is a DPM server. It is DPM backups that are being stored on the MD3000i.
No, I don't believe VSS/VDS providers are installed. Can I download those separately? I don't believe we have a DVD with MDSM. I just downloaded an install file from Dell.
DELL-Kenny K
685 Posts
1955
0
Posted November 4th, 2013 08:00
Here is the link for the resource CD.
Also if you could can we get a DSET report from the server? That may also give us some insight as well being the failover command is coming from the host. Also what version of DPM are you running?
DELL-Kenny K
685 Posts
2008
0
Posted November 5th, 2013 12:00
After looking through everything it looks like the iSCSI initiator is set up incorrectly. It looks like you only have a single IP address setup and the MD has two available. Below is a link that has a step by step about setting up the MD3000i and the MPIO.
http://paulgrevink.wordpress.com/2012/02/21/configure-dell-md3000i-mpio-with-windows-2008-r2/
Let me know how this goes. Also the only connection that you do have keeps dropping itself which normally points to a network issue so may need to look into the switch config as well.