ESX

I’ve spend a lot of time in the past trying to understand the settings for EnableResignature and DisallowSnapshotLUN. It had me confused and dazzled a couple of times. Every now and then I still seem to have trouble to actually understand these settings, after a quick scan through the VCDX Enterprise Study Guide by Peter I decided to write this post and I took the time to get to the bottom of it. I needed this settled once and for all, especially now I start to focus more on BC/DR.

According to the San Config Guide(vi3_35_25_san_cfg.pdf) there are three states:

1. EnableResignature=0, DisallowSnapshotLUN=1 (default)
   In this state, you cannot bring snapshots or replicas of VMFS volumes by the array into the ESX Server host regardless of whether or not the ESX Server has access to the original LUN. LUNs formatted with VMFS must have the same ID for each ESX Server host.
2. EnableResignature=1, (DisallowSnapshotLUN is not relevant)
   In this state, you can safely bring snapshots or replicas of VMFS volumes into the same servers as the original and they are automatically resignatured.
3. EnableResignature=0, DisallowSnapshotLUN=0 (This is similar to ESX Server 2.x behavior.)
   In this state, the ESX Server assumes that it sees only one replica or snapshot of a given LUN and never tries to resignature. This is ideal in a DR scenario where you are bringing a replica of a LUN to a new cluster of ESX Servers, possibly on another site that does not have access to the source LUN. In such a case, the ESX Server uses the replica as if it is the original.

The advanced LVM setting EnableResignature is used for resignaturing a VMFS volume that has been detected with a different LUN ID. So what does the LUN ID has to do with the VMFS volume? The LUN ID is stored in the LVM Header of the volume. The LUN ID is used to check if it’s the same LUN that’s being (re)discovered or a copy of the LUN that’s being presented with a different ID. If this is the case the VMFS volume needs to be resignatured, in other words the UUID will be renewed and the LUN ID will be updated in the LVM header.

UUID, what’s that? Chad Sakac from EMC described it as follows in his post on VMFS resignaturing:

It’s a VMware generated number – the LVM signature aka the UUID (it’s a long hexadecimal number designed to be unique). The signature itself has little to with anything presented by the storage subsystem (Host LUN ID, SCSI device type), but a change in either will cause a VMFS volume to get resigned (the ESX server says “hey I used to have a LUN with this signature, but it’s parameters were different, so I better resign this”).

Like Chad says the UUID has little to do with anything presented by the storage subsystem. A VMFS volume ID aka UUID looks like this:

42263200-74382e04-b9bf-009c06010000

1st part – The COS Time when the file-system was created or re-signatured
2nd part – The TSC Time; an internal time stamp counter kept by the CPU
3rd part – A random number
4th part – The Mac Address of the COS NIC

Like I said before, and this is a common misconception so I will say it again, the LUN ID and the Storage System product ID are stored in the LVM header and not the actual UUID itself. Not that it really matters for the way the process works though.

I think that makes it clear when to use EnableResignature and when not to use it. Use it when you want to access VMFS volumes of which the LUN ID changed for whatever reason. For instance a fail over to a DR Site with different LUN numbering or SAN upgrades which caused changes in LUN numbering.

That leaves DisallowSnapshotLun. I had a hard time figuring out when to set it to “0” and when to leave it at the default setting “1”. But found the following in a VMworld Europe 2008 presentation:

DisallowSnapshotLun: Should be set to “0” if SCSI Inquiry string differs between the two Array’s in order to allow access to datastores.

I googled for “SCSI Inquiry” and I found the following in a document by HP:

The storage system product ID retrieved from the SCSI Inquiry string (Example: HSV210)

In other words, when you’ve got an HP EVA 4000 and an HP EVA 8000 which are mirrored you need to set DisallowSnapshotLun to 0, when a fail-over has occurred. The SCSI Inquiry string would differ because the controllers would be of a different model. (The SCSI Inquiry string also contains the LUN ID by the way.)

When both sites are exactly the same, including LUN ID’s, you don’t need to change this setting. Leave it set to 1. Be absolutely sure that when you set DisallowSnapshotLun to 0 that there’s only 1 “version” of the VMFS volume presented to the host. If for some reason both are presented data corruption can and probably will occur. If you need to present both LUNs at the same time, use EnableResignature instead of DisallowSnapshotLun.

Depending on the way your environment is setup and the method you chose to re-enable a set of LUNs you may need to re-register your VM’s. The only way to avoid this is to use DisallowSnapshotLun and pre-register all VM’s on the secondary VirtualCenter server or use just one VirtualCenter server.

Re-registering can be done with a couple of lines of script on just one ESX box:

for i in `find /vmfs/volumes/ -name "*.vmx" `
do
echo "Registering VM $i"
vmware-cmd -s register $i
done

You can change the EnableResignature or DisallowSnapshotLun setting as follows:

open vCenter
Click on a host
Click on “Configurations” tab
Click on “Advanced Settings”
Go to “LVM”
Change appropriate setting
Click “Ok”
Rescan your HBA’s (Storage Adapters, Rescan)

It’s also possible to use the command line to enable DisallowSnapshotLun or EnableResignature:

echo 0 > /proc/vmware/config/LVM/DisallowSnapshotLUN
echo 1 > /proc/vmware/config/LVM/EnableResignature