disks, service guard, and EMC

[Rick Garland]

Hi all:

HPUX 11.11 runing on rp7410's, MC/SG 11.14.

I have many EMC disks that had been presented to the HPUX systems some time ago, now I need to take them back and present to another host. I am in the process of confirming which disks I can take. What I have been doing is the syminq command on each host, compare the list of hypers with the results of `strings /etc/lvmtab` to ensure that I am not removing anything that has valid data.

Question, what else should I be looking at?

[Geoff Wild · Accepted solution]

Your plan of attack is fine.

After you are done, cleanup the old device files with a:

rmsf -a /dev/dsk/cXtXdX

I have almost same setup as you - except MC/SG 11.15 - which I'm upgrading to 11.16 this Saturday...

Rgds...Geoff

[RAC_1]

What do you mean by valid data?? Do you mean, those who are part of VG?? If that is the case, you can run pvdisplay on each disk, and those which fail on pvdisplay are not used in any of the vgs.

Hope this helps.

Anil

[Steven E. Protter]

I like your plan.

You should be able if necessary to present the disks to multiple hosts at the same time. Since SG when it fails over generally activates the a volume group pointing to shared disk when an application fails over.

You might want to look at this document, in case instance renumbering is something you want to do. Calculations can be easier if a disk available on two nodes has the save driver name/scsi instance number.

http://www2.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&admit=-1335382922+1098972655064+28353475&docId=200000067424466


SEP

[John Poff]

Hi Rick,

It sounds like you have a good handle on it. We've run large EMC arrays and MC/SG here before, and the only other thing I could think of (that I've had to look out for before) is to check the EMC serial number on the disks (that shows up in syminq) to make sure that you aren't grabbing a disk that is seen by another controller.

For example, I've grabbed some old output from a syminq command here:

------------------------------------------------------------------------
DEVICE :VEND :PROD :REV :SER NUM :CAP(kb)
------------------------------------------------------------------------

/dev/rdsk/c9t3d2 :IBM :2105F20 :.114 :01A20122 :13281280

/dev/rdsk/c11t3d2 :IBM :2105F20 :.114 :01A20122 :13281280


You might have a disk in your VG as c9t3d2, and be looking at syminq output thinking that you will be taking c11t3d2, which is actually the same disk presented to two different controllers.

I'm sure you've got a handle on all that kind of stuff, but that's the only thing I can think of to look out for. The syminq output is definitely the way to go.

Good luck!

JP

[Rick Garland]

Hey all:

Many thanks for the responses! Was wanted the confirmation and I got it.

This forum is great!

Thanks again!