09-21-2010 01:36 PM
We're new to Brocade SAN (just converted from Cisco) and we have multiple AIX hosts with Emulex HBAs. All is well except that one HBA is appearing as a target. The other HBA on the same host appears as an initiator as it should. Both HBAs on each of the other 7 hosts appear as initiators. I can't find any differences in configuration or firmware. Any ideas? Also, is it possible to override the "target" status? We have DCFM but not the Enterprise Edition.
09-22-2010 11:38 AM
Did you have checked with the the switch CLI the nameserver entry if it is really a Target?
You can check it with "nsshow -t". Please run "help nsshow" on the CLI and check if the option -t is correct. I have currently no access to the manual.
This will give you the correct status.
Sometimes DCFM produces funny things.
09-22-2010 01:44 PM
Device Type: Physical Target
The other AIX adapters show
Device Type: Physical Unknown(initiator/target)
Is there a switch port setting that could affect this?
I guess we could try reloading the adapter firmware. Maybe a bit flipped.
We forced the adaptor to be an initiator in DCFM but the switch FOS still sees it as a target.
10-21-2010 10:35 PM
Check with the AIX guys if the correct firmware and driver is loaded on the HBA as well as the settings on that particular HBA.
Be aware that an N-Port may behave as an Initiator, Target or both depending on the capabilities of the port. The determination of this is being done during PRLI on the FCP level. It has nothing to do with the fibre-channel layer 0,1,2 and 3 protocol itself. Remember that this is a SCSI thing.
Also check if the correct ODM's are loaded especially in multipath environments.
Hope this helps.
10-22-2010 07:30 AM
One server, two adapters on separate fabrics. First adapter appears as Target, second as Unknown(initiator/target).
Same firmware level 1.90A4 (which is downlevel), same driver (since it's one LPAR), same (writable) settings, generic AIX ODMs.
10-25-2010 05:35 PM
There is too little information to determine why this happens. It's best to open up a supportcall with IBM to verify the HBA settings. This is not a switchport issue.