What is the status of the EMC case on this issue? They have Finisars that they will bring onsite to troubleshoot fabric problems like this. I had two onsite for an issue with my Cisco 9509s for about three months!
So some hosts are logged in and registered to SPA2 just fine, and some hosts are not registered to SPA2 and do not have active paths to that port?
I am assuming you double checked all zones on the switch to ensure the hosts in questions have active zonesets to that port?
For the hosts that do not have working paths to SPA2, what do you see in Navisphere when you expand the Storage Group that the host belongs to and right click on the host and select Connectivity Status? Is there a ~management group listed?
Can you see SPA2 on the hosts through the HBA tools (QLogic SANSurfer or Emulex HBAnywhere)?
Apart from what dynamox and Aran suggested, I like to mention that hopefully you have checked all physical connectivity issue - like the Cable, Switch port, patch panel etc. Check the switch port where SPA2 is connected for any errors. You may also try resetting the switch port.
However, it may be a port issue on the SP as well - so keep EMC support or your CE engaged - SPA may require replacement in that case.
My 2 cents after a long silence in this forum - which in fact my first post in this forum in 2009.
This was very sudden BTW, happened overnight on 12/30 to a handful of hosts. Windows, Celerra, ESX. Logged into SPA0 fine, SPA2 path is dead. (SPB0 ports on same fabric are fine) Isolated to this array. 3CXs in our fabric. Some of these hosts have been fine for years. Celerra is a few months old and had trespassed its control luns a couple of times prior to the rest of these hosts becoming not logged in. Celerra attached to two arrays and only complaining about the one with the SPA2 login issue.
Bouncing the port is in the running, but we're waiting for Saturday when we have an SP on site. Toggling the switch port should cause all of the hosts to log back in but we have to plan for the possibility that no initiators will be able to log back into that port. Support is sending out an SP which we'll replace during our Saturday maintenance period.