ADTX/Ciprico Port mapping problems
by Francois Stark on Apr 23, 2012 at 6:21:15 am
We have four ADTX LH (also called Ciprico Netarray 2215A) arrays, and about 10 controllers. Three of the arrays are in production and one is a test unit. QLogic 5200 and 5600 switches. All 10 clients are running Mac os X.5.8 or X.6.4 with Apple LSI fibre cards, SANMP software. However, my problem lies at a deeper level: Fibre LUN mapping and PWWN and NWWN numbers...
All of the arrays work fine if you connect just one fibre cable to the array, and enable all LUN's on this port. However, if I split the LUN' over two or more ports, then two of the arrays present the same problem: Only one of the ports show up in the system profiler - twice.
So upon closer inspection, I have found that the working arrays indicate the following on the two ports:
NWWN: ...A8 0B 60 49 06
PWWN: ...A8 0B 60 49 06
NWWN: ...A8 0B 40 49 06
PWWN: ...A8 0B 40 49 06
Common issue: the NWWN and PWWN are the same on each port, but differs between port 2 and port 4. This setup works - I assume because the mac clients see different NWWN numbers and the inspects the LUNS seperately.
However, the faulty arrays indicate the following:
NWWN: ...A8 0B 00 41 67
PWWN: ...A8 0B A0 41 67
NWWN: ...A8 0B 00 41 67
PWWN: ...A8 0B 80 41 67
Common issue: The NWWN and PWWN differs on each port, while the NWWN on port 2 is the same as on port 4. As I understood it the NWWN - Node World Wide Number - should be the same for the whole disk array, and the PWWN - Port World Wine Number - should be different for each port on the array. So this is how it is supposed to be set up! However, the mac clients only see port 2, not port 4, and it shows port 2 twice in the system profiler - fibre channel device tree???????
It seems the problem stays with the disk array box, even after changing controllers - so it must be a setting in the LCD controller on each array. This little LCD controller programs the controllers as they are inserted the first time with the info relevant to the array - the disks, arrays, LUNS, LUN mappings, ethernet settings etc.
Anybody with experience on these ADTX arrays? Bart, Chrispy?
It seems that once again - my process of explaining the problem clearly lead me to a solution...
On the array' LCD control panel I found a setting:
Change Config -> Option Setting -> Misc Setting -> Assign WWWN
Which has two options: Common (default) and "Unique By Port".
This setting needs to be changed to "Unique By Port".
I was able to replicate the problem and fix it several times this afternoon on the test array. I'm just waiting for everybody to go home and get off the SAN to implement on the three production arrays.