Bad Things Happen at 2am – How To Resign a VMFS Partition

Well, it started at midnight, a standard hardware upgrade really. Power down the VM’s, shut down the host, and call the DC team to do their thing. Twenty minutes later, got a ring from the DC, that the upgrade had been completed.

I popped open the VC, and well, there was no local storage. It was gone, just… gone. Ok, well it wasn’t really gone, it was just missing from the VI Client, and from /vmfs/volumes, but looking at fdisk –ul:

# fdisk -ul

Disk /dev/sda: 299.4 GB, 299439751168 bytes
255 heads, 63 sectors/track, 36404 cylinders, total 584843264 sectors
Units = sectors of 1 * 512 = 512 bytes

   Device Boot    Start       End    Blocks   Id  System
/dev/sda1   *        63    514079    257008+  83  Linux
/dev/sda2        514080  11004524   5245222+  83  Linux
/dev/sda3      11004525  15197489   2096482+  83  Linux
/dev/sda4      15197490 584830259 284816385    f  Win95 Ext’d (LBA)
/dev/sda5      15197553  17302004   1052226   82  Linux swap
/dev/sda6      17302068 584605349 283651641   fb  Unknown
/dev/sda7     584605413 584830259    112423+  fc  Unknown

Disk /dev/sdb: 299.4 GB, 299439751168 bytes
255 heads, 63 sectors/track, 36404 cylinders, total 584843264 sectors
Units = sectors of 1 * 512 = 512 bytes

It was still there… what to do what to do. With a little help from Toren L. @ vmfaq, I managed to get this sorted:

In most cases this situation can be resolved by doing the following steps:

   1. Choose the ESX Server in viclilent and Click on Configuration / Advanced Settings.
   2. (Optional, depending on environment*) Click on LVM -> LVM.DissallowLVMSnapshot 0 -> LVM.EnableResignature 1
   3. Then click OK to apply settings.
   4. Click on "storage adapters" and click the "rescan" button (upper right). That should let you see the volume again.
   5. Right click the vmhba (under the controller adapter for your machine) and click "rescan".
   6. Then the volume should be visible.
   7. Then when you go to the summary tab, you right click and click "refresh" and you should see your storage1 volume (or whatever you changed it to).
   8. Choose the ESX Server in viclilent and Click on Configuration / Advanced Settings.
   9. Click on LVM -> LVM.DissallowLVMSnapshot 1 -> LVM.EnableResignature 0
  10. Click OK to apply settings.
  11. The name of the visible LUN is now called something similar to snap-5bcf31f1-something. You can now rename it back to what it was called originally. If you have several ESX hosts accessing the LUN you might need to mask out other servers from accessing this LUN in order to rename it.

Anyone else had similar? Care to comment?

* Step number 2 has been corrected – This can cause problems when done on a hypervisor that is part of a cluster, or otherwise has vmotion enabled. It can however, but ignored for most of those cases.