HPUX 10.20 MCSG : Shared DASD Common Disks Between Nodes

[LG Porter]

I have a two K570 class server configured in a MCSG two node cluster. The shared DASD is using a Jamaica box to interace the disks. What is the best method to use when trying to identify the devices files which are used between the nodes? In other words, how can you identify the common disk device file between two node when the hardware paths are different? External Disks are Seagate 17GB disks.

[Alain HONOREZ]

dd if=/dev/rdsk/c0t0d0 of=/dev/null
Like that you can locate with the flashing led yours disks on both systems.

[Sanjay_6 · Accepted solution]

Hi LG,

Only the bus should be different. The disks should be at the same id if seen from both the systems,

say the bus is c0 on node1 and c5 on node2, any disk on the jamaica between these two buses will be seen as ,

node1 ---> node2
c0t1d0 ---> c5t1d0
c0t2d0 ---> c5t2d0
c0t3d0 ---> c5t3d0
c0t4d0 ---> c5t4d0
c0t5d0 ---> c5t5d0

Hope this helps.

Regds

[Stephen Doud]

Hello LG,

A undocumented "feature" of the cmquerycl command only used by the response center to help identify trouble building clusters can be exploited to show which disks are which.
See ITRC knowledge document number UXSGKBAN00000106
"Finding Common Disks Between Hosts or I/O Cards with ServiceGuard"

The technique described uses the command in the following way:

# cmquerycl -l lvm -T 3 -n eon -n ion | grep "added disk"

The technique also prints output fields in a particular order to simplify the report and sorts numerically (based on identified disks) to show related disks.

The technique may require adaptation to your version of ServiceGuard to get the report to look "just right".

-s.