I have a BL460c G7 blade server with OS Windows 2008 R2 SP1. This is a brand new C7000 enclosure, with FlexFabric interconnects. I got my FC switches setup and zoned properly to our Clariion CX4, and can see all the hosts that are assigned FCoE HBAs on both paths in both Navisphere and in HP Virtual Connect Manager.

So I went ahead and created a storage group for a test server, assigned the appropriate host, assigned the LUN to the server. So far so good, log onto server and I can see 4 unknown disks.... No problem, I install MS MPIO, no luck, can't initialize the disks, and the multiple disks don't go away. Still no problem, I install PowerPath version 5.5 reboot. Now I see 3 disks. One is initialized and ready to go, but I still have 2 disks that I can't initialize, can't offline, can't delete. If I right click in storage manager and go to properties I can see that the MS MPIO tab, but I can't make a path active.

I want to get rid of these phantom disks, but so far nothing is working and google searches are showing up some odd results, so obviously I'm not framing my question right. I thought I'd ask here real quick. Does anyone know a quick way to get rid of these unknown disks.

Another question, do I need the MPIO feature installed if I have PowerPath installed? This is my first time installing Windows 2008 R2 in this fashion and I'm not sure if that feature is needed or not right now.

So some more information to add to this. It seems I'm dealing with more of a Windows issue than anything else. I removed the LUN from the server, uninstalled PowerPath completely, removed the MPIO feature from the server, and rebooted twice. Now I am back to the original 4 Unknown Disks (plus the local Disk 0 containing the OS partition of course, which is working fine)

I went to diskpart, I could see all 4 Unknown disks, I selected each disk, ran clean (just in case i'd somehow brought them online previously as GPT and didn't realize it) After a few minutes I was no longer able to see the disks when I ran list disk. However, the disks are still in Disk Management. When I try and offline the disks from Disk Management I get an error: Virtual Disk Manager - The system cannot find the file specified. Accompanied by an error in System Event Logs:

Log Name:      System
Source:        Virtual Disk Service
Date:          6/25/2012 4:02:01 PM
Event ID:      1
Task Category: None
Level:         Error
Keywords:      Classic
User:          N/A
Computer:      hostname.local
Unexpected failure. Error code: 2@02000018
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
    <Provider Name="Virtual Disk Service" />
    <EventID Qualifiers="49664">1</EventID>
    <TimeCreated SystemTime="2012-06-25T21:02:01.000000000Z" />
    <Security />

I feel sure there is a place I can go in the Registry to get rid of these, I just can't recall where and I am loathe to experiement.

So to recap, there are currently no LUNS attached at all, I still have the phantom disks, and I'm getting The system cannot find the file specified from Virtual Disk Manager when I try to take them offline.


  • 1
    Have you zoned these disks to your server on your fiber switch? – jmreicha Jun 25 '12 at 20:00
  • 2
    I'm no EMC expert but is there a CX4 DSM for MPIO? There usually is for most array types to deal with just this kind of thing. – Chopper3 Jun 25 '12 at 20:31
  • @Chopper3 +1. I've had grief in the past trying to get some Dell MD3xxx arrays to work with MPIO. The only thing that fixed it was using the Dell wizard to configure the initiator for me. – Chris McKeown Jun 25 '12 at 23:12
  • Well, there may well be a zoning issue as jmreicha thought earlier now that I've looked further into this. All my servers in this chassis see the same mystery Unknown disks, even those without MPIO or PowerPath loaded. I found I could get rid of them by going into Device Manager - Disk Drives, and uninstalling DGC LUNZ Multi-Path Disk Device, but upon reboot they always re-appear. This may well be a zoning issue, which is a grey area for me but I'll work on it more tomorrow. – Ailbe Jun 25 '12 at 23:22

Thanks JMreicha and Chopper3. I've solved the issue, there was indeed a zoning issue, but even more importantly the fiber was apparently run wrong. I hadn't done a lot (ok, practically no) fiber channel zoning experience before and screwed up the way the fiber switches were connected. I've corrected the issue, fabricA is now connecting just fine over fiber switches 1 and 3 and fabricB is now connecting over fiber switches 2 and 4. All uplink ports on the switches are set to E ports, and all host ports are E and F. All zoning is complete and SQL clusters are being built as I type this out.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.