Check HLU (Host LUN Number) of the LUN. I can imagine that if the address is the same, ESX is confused. Tried a rescan on ESX yet ?

I'm pretty sure it's an ESX thing, since the host hasn't changed in Navisphere.

Also be sure that the ESX servers do not see the LUN at one time (rescan) and then allocate the LUN.
I had recently a problem when deleting a LUN and adding another LUN. I had to do in two steps.

Your ESX was confused

In 2 steps is indeed better. For example: remove it, rescan and add the new one.
Or: add the new one with a different HLU and remove the old one and rescan, then change the HLU and rescan

True! And it was very curious because the new LUN seemed to be correctly seen and with the correct size. But when creating a VMFS partition into it, then the problem began on one node (the others node were OK).

Or simply pull it from one SG first, do a rescan on your host and then put it into another SG and do a new rescan on your host. You don't need engineering mode for that.

If you move a LUN to another Storage group the "Host ID" will change to 0. this will make the ESX server not see it. You have to make a note of it and change it when you move it in Engineering mode.

Thanks guys,

Next time I'll follow those steps, the problem was that after the removal the ESX server recognize the LUN as a snapshot so we follow this WMware steps:
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=9453805

Thanks again for the help guys!!