02-10-2014 10:48 AM
I've a strange issue & couldnt get through the fix yet,requires all your help
Well, I've three Windows Hosts connected to a access gateway in HP C7000 Blade enclosure and AG inturn connected to an Brocade DCX edge switch. As part of our Maintenance activity, we added an extra path for existing zones for all the three servers & enabled the zone config. Immedietly, after enabling the zone config, server paths went offline abruptly & lost disk access.
We checked all the possible scenarios but coulnt find the correct fix for this yet.
Finally, paths came online immedietly after reenabling the F-ports in AG where servers are connected. Anyone has any idea why this has happened? Although we didnt do any path removal from zoning.
Kindly help me to understand by providing a recommended solution.
03-04-2014 01:12 PM
Sorry but i did not fully understand your issue. What do you mean by "we added an extra path for existing zones for all the three servers"?
How many zones do you have related to these 3 hosts?
Did you add a new member to the zones?
03-04-2014 01:41 PM
So you added new targets to the initiators for your servers, enabled this new zone config, and immediately lost all targets for those servers? You talk about re-enabling F ports on your blade access gateway, but not how they were initially disabled.
As far as I understand it, an access gateway is something that will let blades log into the fabric using NPIV. When you are up and running, doing a portshow on the Brocade port that AG is wired to should show all the NPIV WWNs from all the servers using it. I'd be curious to see what that output shows when you reproduce the problem. If they're still logged into the fabric but can't access any disk, I'd try a fcping on the initiator and the target. If they aren't logged in, then there's something wrong with the AG setup.
01-06-2015 07:51 AM - edited 01-06-2015 08:00 AM
Was this ever resolved? I am having a similar issue with some LUNs not being visible from certain ports on Brocade 300s when multiple paths to a target port are available. The strangest thing is that the problem is isolated to the actual SFP port on the Brocade 300 switch, for example moving the Fibre Channel cable to another port on the switch restores access to the LUNs for the affected host HBA. Rebooting any switch in the domain also resolves the problem for a couple weeks, but eventually one or more hosts will lose access to LUNs again, and we have to reboot a Brocade 300 to clear the problem.
04-01-2015 05:51 AM
I had this excactly behavior on my SAN this week.
Core: Brocade 48k (FOS 6.4.3f) Director connected to IBM DS8870 and IBM DS8700 Storages.
HP C7000 Bladecenter with Brocade SW5480 (FOS 6.3.0c) Switch module, but, instead of AG, it is running as Switch.
DS8700 is being decomissioned and we are migrating the luns to the DS8870. After enabling new zones to the DS8870 in the SAN, the Blade serves lost access to all luns.
Looking at the porterrshow noticed that for each RX frame, there were a c3_disc.
portstatsshow said that the c3_disc were er_zone_miss errors
Access became OK after disable and enable the port
04-06-2015 03:14 AM
In my case, found the problem being a bug tracked by defect 284472 where RSCN field were corrupetd during a firmwaredownload to 6.3.0c.
Check how are the rscn with nsshow -r. the SCR field must be non zero. (possible values are 3 and 2, as far as I know). But if you have SCR zeroed, you may having the same problem.