[ale] Suse 9.3 and fiber storage

Damon L. Chesser damon at damtek.com
Sat Apr 16 12:54:29 EDT 2011


On Sat, 2011-04-16 at 11:27 -0400, Andrew Wade wrote:
> So in your migration, I'd zone the existing disks to your RHEL 5/6
> server then use the rescan-scsi-bus.sh script to map in the disks
> through the HBA and from there you can duplicate your setup and map
> the disks to a filesystem.
> 
> 
> But, beware that these are probably not on a clustered filesystem
> (gfs ,etc), so unmount it from the Suse side first before mounting on
> the RHEL side.  You don't want the systems both messing with data on
> the same time on a ext3 or ext4 filesystem.

This I will not be able to do:  The old server with the mounts must
remain intact to insure a recovery is able to be done in the event of a
cut-over failure.  The data will be built new on the new server with the
new mounts.

However, the RHEL specific details you gave will help me out much,
thanks!
> 
> On Sat, Apr 16, 2011 at 11:23 AM, Andrew Wade <andrewiwade at gmail.com>
> wrote:
>         I work on Fiber Channel connected SAN disks on RHEL 4, and 5.
>         
>         
>         First you need to enable multipath
>         
>         
>         service multipathd start
>         chkconfig multipath on
>         
>         
>         Next, you need to install (for RHEL 5) sg3 ultils.  Make sure
>         you're on the right Channel in Red Hat Satellite to get those
>         packages.
>         
>         
>         yum install sg3*
>         
>         
>         Next, zone the LUNS to your WWN of your HBAs on your desired
>         servers.
>         
>         
>         Then to discover the Luns that were zoned to your sever, look
>         in /usr/bin/rescan-scsi-bus.sh
>         
>         
>         
>         
>         
>         
>         If you have RHEL 4, I'd install the qlogic san surfer utils to
>         use their script to scan in the luns.   In RHEL 5, native sg3
>         utils work great and you don't have to worry about patching
>         your initrd, etc.
>         
>         
>         
>         
>         Andrew Wade
>         RHCE
>         This script with no options scans the hba's for new luns and
>         maps them in.
>         
>         
>         Before running the script, you should make a backup copy of
>           /var/lib/multipath/bindings
>         
>         
>         Then run the rescan-scsi-bus.sh script.
>         
>         
>         Next do a diff on /var/lib/multipath/bindings and
>         your /var/lib/multipath/bindings.bk
>         
>         
>         You'll see the new LUN that got added.
>         
>         
>         multipath -ll    will confirm all the luns you've added
>         
>         
>         Now, for custom aliases, you go back to your bindings file and
>         delete the entry it made called mpath0  and the uid of the
>         disk.
>         
>         
>         Then edit your multipath.conf to use the uid of the disk
>         (which you see in multipath -ll ) and give it the alias you
>         want ie: oracle_asm01 , etc.
>         
>         
>         Then you run multpath -v3  to reread your multipath.conf and
>         rebuild the mulitpath configuraton with your new disk alias
>         (instead of the generic mpath0, mpath1, etc.)
>         
>         
>         Next, look at /etc/multipath.conf
>          (or /etc/multipath/multipath.conf  I'd have to look).   Here
>         you can define your aliases if you have any.
>         
>         
>         
>         On Fri, Apr 15, 2011 at 7:50 PM, Greg Freemyer
>         <greg.freemyer at gmail.com> wrote:
>                 For figuring out what you have:
>                 
>                 You're getting too complex for what seems a simple
>                 job.
>                 
>                 FC volumes are scsi normally, so /dev/sdb, etc is
>                 likely the drives.
>                 
>                 Just like a physical drive, a FC volume can be used in
>                 whole or partitioned.
>                 
>                 To get the full unpartitioned volume size, look
>                 in /sys/block/sdb/...
>                  (You can also just call df.)
>                 
>                 You should be able to get partition info
>                 from /proc/partitions
>                 
>                 You should see all of your mount points the
>                 traditional way.  ie. Look
>                 in /etc/fstab and/or run mount.
>                 
>                 The key thing is FC drives fit into the normal scheme
>                 at the level you
>                 are talking about.
>                 
>                 You will have a little more fun setting up the new
>                 environment and
>                 mounting the volumes.
>                 
>                 Also, you can't tell how the raid setup is done from
>                 the basic linux
>                 side.  (There may be management software that tells
>                 you, but that will
>                 be a separate thing.  Likely the storage guys have
>                 that info and you
>                 don't.  Trouble is they need to know more detail than
>                 just /dev/sdb
>                 nomenclature to know which volume you are talking
>                 about on their end.)
>                 
>                 Greg
>                 
>                 
>                 
>                 
>                 
>                 On Fri, Apr 15, 2011 at 4:51 PM, Damon L. Chesser
>                 <damon at damtek.com> wrote:
>                 > I have a bunch of Suse 9.3 servers with various apps
>                 that need to be
>                 > migrated to RHEL 5 or 6.
>                 >
>                 > I have back end SANs attached via QLogic hbas.
>                 >
>                 > How do I verify how the attached storage is mounted
>                 (ie, this mount is
>                 > remote via the hba).
>                 >
>                 > There is no /etc/mulitpath.conf
>                 >
>                 > What I am looking for is a way I can get info and
>                 make a "map" that I
>                 > can duplicate on the new OS.  The new storage will
>                 be entirely new
>                 > partitions/luns on completely different LUNs, but
>                 the "structure" might
>                 > need to be the same, ie: /somemount is
>                 17G /somemount2 is 15G etc.
>                 >
>                 > /dev/disk/by-* has by-id and by-uuid and by-path.
>                 >
>                 > I know this is both rather simple and broad, but I
>                 have zero fiber/HBA
>                 > experience and it would appear I don't know the
>                 proper search terms to
>                 > google.
>                 >
>                 > If it matters the back end is (old) is a Hitachi and
>                 I don't know the
>                 > front end.  I will not be tasked with slicing up the
>                 LUNs, but reporting
>                 > what sizes I need them to be, then mounting the
>                 partitions with the
>                 > proper mount points on the new OS.
>                 > --
>                 > Damon
>                 > damon at damtek.com
>                 >
>                 > _______________________________________________
>                 > Ale mailing list
>                 > Ale at ale.org
>                 > http://mail.ale.org/mailman/listinfo/ale
>                 > See JOBS, ANNOUNCE and SCHOOLS lists at
>                 > http://mail.ale.org/mailman/listinfo
>                 >
>                 
>                 
>                 
>                 
>                 --
>                 Greg Freemyer
>                 Head of EDD Tape Extraction and Processing team
>                 Litigation Triage Solutions Specialist
>                 http://www.linkedin.com/in/gregfreemyer
>                 CNN/TruTV Aired Forensic Imaging Demo -
>                 
>                 http://insession.blogs.cnn.com/2010/03/23/how-computer-evidence-gets-retrieved/
>                 
>                 The Norcross Group
>                 The Intersection of Evidence & Technology
>                 http://www.norcrossgroup.com
>                 
>                 
>                 _______________________________________________
>                 Ale mailing list
>                 Ale at ale.org
>                 http://mail.ale.org/mailman/listinfo/ale
>                 See JOBS, ANNOUNCE and SCHOOLS lists at
>                 http://mail.ale.org/mailman/listinfo
>                 
>         
>         
> 
> 
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo

-- 
Damon
damon at damtek.com



More information about the Ale mailing list