Re: AIX/SAN Storage
From: N N (delalist_at_HOTMAIL.COM)
Date: 05/04/05
- Previous message: Michael Cheselka: "Re: Problem with var directory being full!"
- In reply to: jeff barratt-mccartney: "Re: AIX/SAN Storage"
- Next in thread: Damir Delija: "Re: AIX/SAN Storage"
- Reply: Damir Delija: "Re: AIX/SAN Storage"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 4 May 2005 15:57:55 -0500 To: aix-l@Princeton.EDU
Thanks for all the feedback...
My thoughts....
Unless a particular vendor does it differently part of the FC_ID is the
DOMAIN ID of the switch and another part is the PORT IDENTICATION # (AREA).
Correct me if I am wrong but if you move ports the FC_ID will always change
as the PORT IDENTIFICATION (AREA) portion of the FC_ID changes. Anyone
disagree?
I can understand how this WOULD not cause new hdisk if you had the box down
or rebooted it. I wonder if you...
- took the paths offline in powerpath
- removed the paths from powerpath config
- rmdev -l <hdisk for particular path that is being moved>
- switched the ports
- mkdev -l <hdisk for particular path that is being moved>
- add back to powerpath config and turn paths on
Basically, maybe bringing the devices associated with the path down BEFORE
making the change would prevent new hdisks from being created. That would
be compared to just moving the cable on the fly and letting powerpath handle
all of the path failure detection and reactivation.
Sounds like the people who repsonded have had different results. There has
to be some factor that caused new hdisks to get created one time and not
another.
The reason I thought maybe the storage type makes a difference is because
depending on the driver, disk type and presentation method the attributes
under lsattr and lscfg for a LUN (hdisk) can be different. (Example would
be Shark LUNS vs SVC Virtualized LUNS) If one of the attributes changed when
moving ports that may cause the new hdisks to get created. Since I had
heard of this happening primarily with EMC storage I thought maybe the LUN
definition in AIX is different for IBM disks vs. EMC disks. Has anybody had
this happen with IBM storage or were the references strictly for EMC
storage?
>From: jeff barratt-mccartney <jbarratt@COMPSAT.COM>
>Reply-To: IBM AIX Discussion List <aix-l@Princeton.EDU>
>To: aix-l@Princeton.EDU
>Subject: Re: AIX/SAN Storage
>Date: Wed, 4 May 2005 08:35:38 -0400
>
>neither of these options refer to FC_ID.nor does the storage device make a
>significant difference.
>
>hard and soft zoning, and VLANing have no affect on FC_ID. Since you
>mention
>VLAN, then you have cisco switches, which allows you to maintain a
>persistent FCID, making this argument a moot point in that environment.
>
>my experience is the same as jonathon's, there seems to be no hard and fast
>guarantee that FC_ID will not change when moving across ports/switches.
>Does
>anyone out there have conflicting information?
>
>
>-----Original Message-----
>From: IBM AIX Discussion List [mailto:aix-l@Princeton.EDU]On Behalf Of
>Hass, Harold
>Sent: Tuesday, May 03, 2005 6:38 PM
>To: aix-l@Princeton.EDU
>Subject: Re: AIX/SAN Storage
>
>
>You're right, it depends. On EMC, you can do WWN or Port Masking. With
>WWN masking, as long as the HBA WWN doesn't change, and you haven't
>VLAN'ed the switch, you should be OK. With Port Masking, you have to
>delete and redo the masking.
>
>Port Masking makes it easier to change HBAs; WWN Masking makes it easier
>to swap ports. Pick your poison.
>
>H. Hass
>
>-----Original Message-----
>From: IBM AIX Discussion List [mailto:aix-l@Princeton.EDU] On Behalf Of
>Jonathan Fosburgh
>Sent: Tuesday, May 03, 2005 1:45 PM
>To: aix-l@Princeton.EDU
>Subject: Re: AIX/SAN Storage
>
>On Tuesday 03 May 2005 13:53, N N wrote:
> > A single HBA AIX SAN connected system plugged into port #1 of a Fibre
> > Switch. If the fiber connection is moved from port #1 to port #2 the
> > Fiber channel D_ID (destination ID) changes. I have heard that at
> > times AIX will see the existing SAN LUNS as new hdisk devices (when
> > cfgmgr is run) because of the different D_ID. I have moved SAN ports
> > for AIX systems plenty of times and have never seen this when going to
>
> > a Shark or FAStT. Has anyone experienced this? EMC or IBM storage?
> >
>My experience is that IT DEPENDS(tm). If the FCID of the port changes
>then AIX will probably lose track of the devices. But, I have seen that
>turning off the system before making the move will sometimes allow the
>change to be seen and the devices will come back. Note that this has
>been inconsistent for me. Far better is to enable persistant FCIDs on
>the switch if supported, and/or enable dynamic tracking on the HBA if it
>is supported. YMMV though, persistant FCIDs is your best bet.
>
>--
>Jonathan Fosburgh
>AIX and Storage Administrator
>UT MD Anderson Cancer Center
>Houston, TX
_________________________________________________________________
Don’t just search. Find. Check out the new MSN Search!
http://search.msn.click-url.com/go/onm00200636ave/direct/01/
- Previous message: Michael Cheselka: "Re: Problem with var directory being full!"
- In reply to: jeff barratt-mccartney: "Re: AIX/SAN Storage"
- Next in thread: Damir Delija: "Re: AIX/SAN Storage"
- Reply: Damir Delija: "Re: AIX/SAN Storage"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
