MCSG 10.10: Identification Of SCSI Disks Between Nodes

[LG Porter]

HPUX 10.20 - I have a two node cluster presently running MCSG 10.10. All the shared disks are HP Seagate disks. When the primary node fails over to the secondary node, the hardware path changes. What is the best method for determining the disk hardware path and device files between the two nodes? Thank You

[A. Clay Stephenson · Accepted solution]

Generally a combination of cmquerycl -v and ioscan -C disk -fn on each node.

[PIYUSH D. PATEL]

Hi,

The device files are bound to change if they are installed on different controllers in a different manner.

NodeA may have device file /dev/dsk/c1t2d0
NodeB may have device file /dev/dsk/c2t2d0 for the same disk.

Using ioscan -fnC disk and cmquercl -v you can find out the hardware path and device files.

During planning of a MCSG cluster, the hardware paths and devices have to be taken into consideration.

Piyush

[Christopher McCray_1]

Hello,

I would use the cmscancl command along with ioscan which should give you the output you need.

Hope this helps

Chris

[linuxfan]

Hi,

One of the easiest ways i have found is to use inquiry utility from emc. It works for disks other than emc.

you can get inq from
ftp://ftp.emc.com/pub/symm3000/inquiry/latest/

And it is available for all kinds of platforms.

-HTH
Ramesh

[melvyn burnard]

We use an unsupporetd script, written by one of my colleagues. I attach it as it is really useful.

[Frederic Sevestre]

Hi,

If you want to share volume group for the cluster, you can use vgimport with the -s (shared) otion.

Create the vg on the first node, then create a mapfile usin vgexport :

# vgexport -v -p -s -m mapfile

then send the mapfile on the other node and import the vg :

# mkdir /dev/
# mknode /dev//group c 64 0x....
# vgimport -v -s -m mapfile

Then you can check the difference for the hardware path using vgdisplay on both nodes.

Fr??d??ric

[Stephen Doud]

Hi LG,

At SG 10.10 and later, an undocumented option can be exploited to find common disks between nodes: -T

Example:

cmquerycl -T 3 -n | grep "added disk"

Debug (rong) added disk at /dev/dsk/c0t1d0 (ID 418374281 1022698686) 8.1.0 scsi3
Debug (rong) added disk at /dev/dsk/c0t1d1 (ID 418374281 1022783680) 8.1.1 scsi3
Debug (rong) added disk at /dev/dsk/c1t5d0 (ID 418374281 917835374) 52.5.0 scsi1
Debug (rong) added disk at /dev/dsk/c1t6d0 (ID 418374281 917835377) 52.6.0 scsi1

Adding the other node to the cmquerycl command, and using the "awk" and "sort" commands, the output can be prettied up:
# cmquerycl -T 3 -n rong | grep "added disk" | sort -n -k 9,9 | awk '{print $2,$10, $6,$9}'

(rong) 8.1.0 /dev/dsk/c0t1d0 1022698686) \the same disk
(rite) 12.1.0 /dev/dsk/c1t1d0 1022698686) /
(rong) 8.1.1 /dev/dsk/c0t1d1 1022783680) \the same disk
(rong) 12.1.1 /dev/dsk/c1t1d1 1022783680) /

-s.