Confusion with Iscsi target no.

[naveenvm]

With 2-3 of my ESX servers I am experiencing this confusion. vmhba32 is the device for the software iscsi initiator I m using. In the configuration tab-> storage adaptors, it is showing just two targets connected i.e 0,1. Also inside /var/lib/iscsi/vmkbindings it is showing me just two targets connected.

But few of the vmslice's datastore devices are showing me tagets such as vmhba32:2:24 & vmhba32:3:16 etc.

How is it possible. Any ideas?

[weinstein5]

Welcome to the forums - I think I understand what you are asking - first let me explain what vmhabA:B:C:D - A is the HBA adapter (this can be a SCSI cars, RAID controller, FC HBA, iSCSI Hardware HBA and iSCSI software HBA), B - is the Target ID, C - LUN and D- partition on the LUN - so if I understand you question is why do you have vmhba0:x:y:z, vmhba1:x:y:z and vmhba32:x:y:z - you might be seeing other storage adapter cards -

[mike_laspina]

Hello,

What you are seeing is the Controller:Target:LUN:Partition format or commonly seen as C:T:L:P.

Generally the local ESX host will in some way enumerate the Controllers with VMware's vmhba[Controller#]:[TargetPath]

and then the :[Partition] should be the same accross hosts when it is working correctly.

An example of this from a typical host would be if you have an LSI and Adaptec controller locally on the PCI bus it would result in vmhba0:0:x:x and vmhba1:0:x:x

Add the Software iSCSI stack and VMware pads it with 32 resulting in vmhba32:0:x:x for the first target found.

If more targets are found VMware will add a to the target value e.g. vmhba32:1:x:x

If the target is removed it will still add on to what was recorded in the binding file so you could have vmhba32:2:x:x after some targets were removed unless you cleared the binding files.

Hope that helps.

[naveenvm]

Thanks Mike, That looks to be the right reason. As I understand, if I removed one target(from target 0 & 1) at some point of time, say target 0 and agin add the same target it may be visible as target 2.

But I notice that there is a disperancy in the VI client over this, as at the place i.e configuration - > storage adaptor - > it is showing me all luns attached via. target 0 and 1, but if I select a particular vm-slice, the associated lun(device format), shows me target 2. Still wondering over the reason for the disperancy..........

[mike_laspina · Accepted solution]

Yes. That's correct and it does this for a good reason. For example if it were to discover a CTLP and it assigned it to target 1, then the shared storage went down for a period of time. Now a new shared store comes on line. If it were to use the same CTLP on a newly found UUID it would no longer work with the old one when it came back online.

N.B. Points are a great way to say thanks.

[naveenvm]

Agreed on all points with Mike, and confusion is now gone...................BUT

How will I come to know that a specific device with naming i.e vmhba32:2:32:0, is from which target when I have only two targets connected as target 0 and target 1. The file /var/lib/iscsi/vmkbindings also shows me two targets - 0,1.

How will I now know that device vmhba32:2:32:0 is from which of the 2 targets from 0,1?

Any new way for that????????????