12-27-2010 07:22 AM
I'm new to SAN technology especially with fabric OS.
Here is my problem. We have two SAN switches sw300 with all the configuration running properly.
A week ago we get an alert for a faulty fan so we order a new switch. We also received some new license for the new switch.
We plug the new switch create all zoning from scratch (we didn't use configdownload because fabric OS is not the same with the old one).
It worked fine for about an hour and the ports change to no_sync. We plug the old switch (the one with the faulty fan) everything is ok (exept the fan )
They send us another switch we create the zones but we do not see the alias names on the gui only the WWNs. The switch works fine but we get a warning on our RPAs complaining that some partitions use single path (all partitions, disk arrays, RPAs are connected on two switches for redundancy).
The only problem i can think of is the fabric OS firmware. Maybe they are not compatible.
The fabric OS version of the first two switches that worked fine are both V6.1.1a
now we have one switch with fabric OS V6.1.1a and the new one with V6.3.1a. From the compatibility list i cannot see any problems between these two versions. I read that V6.1.1a has a lot of bugs and i should upgrade but i don't want to risk it since we do not have multipath due to our problem.
Can you please advice me on how to proceed? Is it because of the versions OR something else you can help me with?
12-28-2010 10:47 PM
Did you check the domain ids of your switches ? sounds like there could be a issues with domain id mis matches hence your ports going into a funky state.
Running different versions of fabric OS should not cause these types of issues , its best practice to have all the same OS level's but it won't hurt having some versions slightly different.
BTW you should not need to recreate your alias/zones from scratch , that should all flow down from your other switch in the fabric.
12-30-2010 12:22 AM
Thanks for your reply.
Sorry if i didn't explain my topology correctly. The switches are not in the same fabric. They are totally indepented paths. On each switch we have connected one of the two HBAs of each server, same for the RPAs and the SPs of the disk array. This way all servers can access the disks from two indepented paths. In such a topology we do not have to worry about domainID (is that correct? Does it matter if the domain IDs are the same or not?).
zoning: ON (CFG_SWITCH_23)
Index Port Address Media Speed State Proto
0 0 011700 id N4 No_Light FC
1 1 011500 id N4 No_Light FC
2 2 011300 id N2 Online FC F-Port 50:06:01:69:30:20:92:e8
3 3 011100 id N2 Online FC F-Port 50:06:01:61:30:20:92:e8
4 4 011600 id N4 No_Light FC
5 5 011400 id N4 No_Light FC
6 6 011200 id N4 No_Light FC
7 7 011000 id N4 No_Light FC
8 8 010f00 id N4 Online FC F-Port 50:01:24:82:01:ab:a0:92
9 9 010d00 id N4 Online FC F-Port 50:01:24:82:01:ab:cb:8a
10 10 010b00 id N4 No_Light FC
11 11 010900 id N4 No_Light FC
12 12 010e00 id N4 Online FC F-Port 50:01:24:82:01:ab:38:92
13 13 010c00 id N4 Online FC F-Port 50:01:24:82:01:ab:21:8a
14 14 010a00 id N4 No_Light FC
15 15 010800 id N4 No_Light FC
16 16 010700 id N2 Online FC F-Port 10:00:00:00:c9:46:92:05
17 17 010500 id N2 Online FC F-Port 10:00:00:00:c9:4e:39:2a
18 18 010300 id N2 Online FC F-Port 10:00:00:00:c9:4f:29:ad
19 19 010100 id N4 No_Light FC
20 20 010600 id N4 No_Light FC
21 21 010400 id N2 Online FC F-Port 10:00:00:00:c9:4e:49:75
22 22 010200 id N2 Online FC F-Port 10:00:00:00:c9:46:3c:b1
23 23 010000 id N2 Online FC F-Port 10:00:00:00:c9:45:01:48
Sorry if i'm not very clear with my discription but my kwnolege on this subject are limited.
01-03-2011 05:39 AM
If your servers are old AIX servers then you have to run the rmdev on the failed path devices and recreate the devices.
Some AIX versions stored the local SAN port IDs and the storage port IDs. If you change the DomainID of a switch it will not find the old way and the pathes are declared as dead. AIX is capable of dynamic path tracking.
01-06-2011 11:27 PM
Thanks for your help. You where absolutely right!!! The problem was on the hosts (AIX) and not on the switch. We tried cfgmgr and the paths came up alive. We also reboot the servers and the warnings were gone.
Thanks again. I appreciate your help.