Start a Conversation

Unsolved

This post is more than 5 years old

2 Intern

 • 

5.7K Posts

1637

November 25th, 2009 03:00

fcid change / HPUX problems

Hello friends,

I have an HP-UX attached via Cisco MDS 9124 switches to a Clariion. The VSAN and domain numbers were changed to better fit in the VSAN and domain numbering we're using on other locations (unique numbers). Powerpath works fine and access to the LUN's remained, but the ctd numbers in unix changed because of the fcid change. I happen to know what the old fcid is, so I was wondering if the ctd numbers will change back to what they were when I change the fcid back to what they were.

Check check, double check: can I for example change my fcid from 0x370800 back to 0x0a0300 for that particular host ? Do I also need to change the fcid of the attached Clariion or is fixating the host wwn enough ? Will the HP-UX host now see the old ctd numbers ?

When I check the switch and look at "fcdomain fcid database",there's "dynamic" behind each and every wwn. Will it be enough to simply get rid of the "dynamic" part ?

2 Intern

 • 

20.4K Posts

November 25th, 2009 06:00

if you look at this description, hpux disk path contains domain id in the string so i am not sure if simply changing fcid would help you. Would it be easier at this point to re-import volume groups ?  You probably have discovered this already that Cisco docs say to enable static domains and persistend fcid for aix/hpux.

http://docs.hp.com/en/AB378-96001/ch01s12.html

Let us know what you figure out

2 Intern

 • 

5.7K Posts

November 25th, 2009 07:00

Thanks, but the UX was attached to the Clariion some time ago (not at our main site) and we didn't know at the time that we needed to set the fcid to static. So besides the fcid I also need to set the domain id back to what it was before ?

Anyway, by the end of next week we'll shutdown all hosts, including this UX, and after the planned maintenance (including the fcid change) we'll boot UX again and verify if the ctd's are back to normal. If not the UX guy will have to fix it on his side with vgimport or whatever magic he needs to get this done. But Powerpath works great ! Although the software / OS addresses the wrong ctd, PPath redirects it and the LUN is accessable

I'll keep you up2date. If anyone has other hints, I'm open for suggestions.

14 Posts

November 27th, 2009 02:00

Hi,

as dynamox already metioned, you should enable Persistant FCID, otherwise everytime the HBA logs out and in again, it will get a new FCID.

Cisco doesn't assign FCID on Hardware Port Basis. Its more like, first come , first served :-)

A part of the FCID is the Switch Domain ID. You sould set all Domain ID's to Static and use only uniqe Domain ID's (maybe you like to Merge Switches to a Fabric in the future). Be Careful, you have to disable the VSAN / Switch when you change the Domain ID.

You could assign a predetermined FCID to a Host.

Please check out the the Cisco MDScookbook from the Cisco Website.

Ciao, Marco

14 Posts

December 9th, 2009 12:00

Hi,

the Domain ID is Part of the FCID.

DomainID     F_Port on switch     AL_PA
    01                     15                    00    

Ciao, Marco

2 Intern

 • 

5.7K Posts

December 9th, 2009 12:00

I tried to set the FCID fixed to the number it had before, bus since the domain id has changed I wasn't able to do so. It seems the FCID should always match the domain id. Therefore I changed the domain id back to what it was and fixed the FCID fot the wwpn's of the HPUX. I booted the host, but it simply got 4 new paths to each LUN, so the whole action didn't work out the way we thought it would. Mr UNIX had to do some HPUX magic and get rid of the wrong ctd's.

By the way: we had PPath installed, so access to each LUN was till working fine, although errors showed up in some HPUX logging here and there. It was all cosmetic after all

2 Intern

 • 

5.7K Posts

December 10th, 2009 01:00

Yeah yeah, I know, but since you can set it manually, I thought you could choose an FCID outside the domain of the switch's domain id. And unfortunately that wasn't the case.

2 Intern

 • 

1.3K Posts

December 10th, 2009 01:00

Once we had a situation as a consequnce of a code upgrade(9.6.x) where FCID got changed ( Not appeared immediatly, but only after the NEXT FLOGI ) as NPIV got enabled as part of upgrade.

That effected all of our hosts , but only on TWO paths. We could survie with rest two paths. Later FCID changed was reverted by disabling NPIV and we broguht all the old paths back. Hosts were having HP-UX and Windows OS. But we had to do some extra work on ASM(Oracle cluster) nodes to k

No Events found!

Top