Unsolved
This post is more than 5 years old
115 Posts
0
1000
Issue assigning storage to Windows 2012 cluster from VNX 5500
Hello,
I have a 3 node Windows 2012 cluster node1, node2, node3. Initially I didn't install Powerpath on any of them but 2 days back installed Powerpath 5.7 on node 1 and node 3. After installation, luns which were seen previously by the host were not seen atall. We rebooted node 1 and node 3 after which luns were seen but after a short amount of time those luns were lost again.
During all this node 2 never lost access to the lun.
I assigned a new lun to the cluster which was seen by node 2 instantly but same issue with nodes 1 and 3. Thinking that this might be caused by the powerpath installation, we removed powerpath from node 3, rebooted, had server admin reconfigure MPIO and still see the issue of luns being "dropped".
The thing that amazes me the most is that this is happening only with storage provided by VNX 5500 (flare version 32 patch 201). Any storage provided from older CX4 or CX3-80 has absolutely no problem.
Has anyone seen this kind of behavior? I have an SR open with both VNX and powerpath teams but was hoping for some pointers here.
Thanks,
Sri
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
March 28th, 2013 12:00
why don't you configure PowerPath or MPIO on each node in the cluster ? If you are presenting a LUN through multiple paths you must use some kind of multipathing software or it's corruption waiting to happen.
BlockStorage
2 Posts
0
March 28th, 2013 16:00
I would start by looking at the HBA firmware/BIOS and drivers. Find the latest Windows 2012 compatible versions from the HBA manufacturers web site and update.
kelleg
4.5K Posts
0
March 29th, 2013 15:00
If you do check for HBA firmware, remember to look for the EMC section on the HBA vendors web site - those are the approved and tested drivers, not the generic drivers.
glen
Sridhar246
115 Posts
0
April 15th, 2013 13:00
Hello everyone, sorry for the delayed response. I have been working on couple of projects which were due in past 15 days.
I have installed powerpath 5.7 on all the cluster nodes. I have verified/updated hba firmware to 9.1.10.28 which is the latest driver version as per qlogic website that is approved by EMC. Also connected these hosts to brocade switches with firmware 7.0.2(a).
The issue is seen when a VM with direct attached storage (via NPIV) is live migrated to a host. In this case, the host is losing 2 out of 4 paths but if that same VM is not on a host then that host can see all the paths.
For example, node1 is sitting on host-A. in a 3 node cluster and is live migrated to Host-B. In this case Host-B which was able to see all paths from VNX suddenly loses 2 paths. If Host-A is rebooted it regains all the paths that it was not seeing before but untill a reboot those paths are down.
In all of this the VM itself is never losing connection to its storage.
VMs and Hosts are in different storage group and see different luns. VMs and hosts both are zoned/masked to the same 4 storage ports (2 on each SP ofc), not sure if that is the underlying issue.
Windows Admin claims that he is doing everything as per Microsoft's guidelines in creating vSAN and vHBAs on Hyper-V hosts.
At this point I am kind of lost...