LUN0 message in iSCSI v6 Beta

G'day All,
Just starting our testing with V6 Beta Upd3, and I noticed that there is now a message when you create a iSCSI target:

Info
Please note that in order to access iSCSI-enabled data from an initiator, the target needs to have a LUN0, otherwise the data in all other LUNs will be inaccessible. The data will also be inaccessible if you select an inactive snapshot or a destination volume (volume replication) as LUN 0.

Now, the second part is fairly straigntforward, but I don't quite understand about the LUN0 issue.
In the past we have chosen to NOT use LUN0, becuase it seems that our ESX servers liked finding the DSS better that way.
And http://kb.open-e.com/entry/134/, specifically advocates not using LUN0 for Virtualiron.

So, can someone please clarify what the above message actually means?
Does it mean that every Target should have a LUN0? OR just one Target needs a LUN0?

Rgds Ben

I find that very confusing, too. From my own experience there're several iSCSI initiators that will not see any LUN of a target if LUN0 of that particular target is present.
GlobalSAN and Atto's XTendSAN are two of these on OS X.

On the other hand, VMWare seems to have trouble, if moren than one LUN0 on any target is available. This is why all the Thecus NAS' will automatically increase the LUNs on each configured iSCSI volume, even if these volume are on seperate SCSI IDs.

The problem is, that neither VMWare, nor some iSCSI initiators really stick to the RFCs here, since it's absolutely nonsense to demand that each LUN is only present once throughout any SCSI ID (VMWare) and neither it's okay that some iSCSI initiators will stop searching the remaining LUNs of a SCSI ID if LUN0 is not present.

There are however at least some iSCSI initiators that won't care if LUN0 isn't present and that will scan for the remaining LUNs nevertheless. The open-iscsi initiator at least does this and afaik, the Windows initiator does this as well.

Cheers,
budy

Quote:

---

Originally Posted by budy

From my own experience there're several iSCSI initiators that will not see any LUN of a target if LUN0 of that particular target is present.

---

Oops… of course thats should read "isn't present".

Btw, it would be really nice if one could edit one's post afterwards…:D

Cheers,
Budy

G'day Budy,
Thanks for the confirmation. We have seen the same thing here, VMware likes different LUNs, while Windows didn't care.
Took us hours to work it out when first trying to get iSCSI working, ie created a second LUN and could get the MS Initiator to see it (from within a VM too!) while ESX refused to play ball until we incremented the LUN.

Our V6Beta testing so far has shown that our previous methods (Not using 0 or reusing LUNS) works, so we will stick with that for the moment. although it may become an issue with Multipathing, Replication or Bonding, as we have not tried any of these yet.

RE Editing post, yeah I agree, it's a real pain when you write something at 1AM, and then read the mistakes at 9AM :D

Rgds Ben

Hey beng

here is a link about not using tha same lun numbers globaly with vmware
http://communities.vmware.com/message/701181#701181