Lun mapping - duplicate LUNid in ESXi

[dpils]

Hi

We have NetApp array presenting iSCSI  to ESXi 5.0u1 servers.

I am trying to see if I can identify iSCSI devices presented to ESXi server  as LUNs created on NetApp. Until now I was under impression that this mapping should be one to one for our setup but this does not seem to be a case:

This is what I see in devices list on ESXi server. As you can see, the LUNid is the same, but size is different.

From NetApp array, only one LUN (with Size 64 GB)  with LUN ID 4 is presented. First Initiator Group is not used.

Does it mean that ESXi server LUNID mapping is out of sync with NetApp array?

How would I be able to identify LUNs then - I thought that LUNid would be a unique identifier but now I am at loss.

We do use RDMs in virtual mode if this helps to clarify the setup.

Thank you!

[kermic · Accepted solution]

Both LUNs have unique naa ids. This makes the host think that those actually are different SCSI devices.

One of your LUNs has a runtime name of vmhba40:C0:T0:L4 while the other one has vmhba40:C0:T1:L4, they are behind different targets.

I would first of all check all initiator groups that have the affected host's iqn included and then check the LUN list on NetApp, looking for an 8GB LUN mapped to any of those groups.

If you have VSC integrated with vCenter, you might find it more comfortable for the task above (I would :))

To make sure you're on the right page at some point (i.e. before removing the LUN mapping or the LUN itself, in case you find it neccessary) you might need to compare LUN serial number with NAA ID reported by ESXi.

The only way I've found so far is NetApp CLI: lun serial -x <path to lun>

This returns the LUN hex serial number that in fact matches NAA ID seen by the host. The downside here is that you need to specify a path for each LUN you want to check which might be a considerable effort if you have 100 LUNs to check.

If anyone here knows a better way, please share!

WBR

Imants

[kermic]

And the essential question, just in case. Are those LUNs still there after rescan?

WBR

Imants

[jdptechnc]

Is your netapp array in an HA configuration?  Maybe you have your host's initiators in multiple igroups, each of which would have a LUN mapped to id 0?

[kermic]

If this is suspected to be the case, you can try to identify the target details by going to host configuration in vSphere client > configuration > storage > devices

click on the LUN in question and click on Manage Paths... link

this should show you the target identification info for each path and might make your quest a bit easier

WBR

Imants

[dpils]

Thank you very much for your tip about 'lun serial -x' !

This helped me to drill down and find out what was wrong:

I have missed the fact that this ESX uses LUNs presented by NetApp and vfiler on the same physical NetApp.

This, of course , results in situation where LUNs presented  from NetApp and vfiler have independend LUN Ids.