Snapshot disk change device file in /dev/dsk after a reboot

[Larry J. Griswold]

We are running HP-UX 11.11v1 on rp4440 server, connected to a EVA3000 SAN. The following lvol is our Oracle database disk presented by the SAN

/dev/vg02/lvol1 102400000 1569352 100043296 2% /oradat

The snapshot disk presented from the SAN for the oracle database disk change everytime we reboot the system. Example:

Previous to reboot the snapshot disk was

/dev/dsk/c10t0d5

After reboot the snapshot disk is now:

/dev/dsk/c10t0d6

I have to change our create/delete shell scripts for the presenting of the snapshot each night after every reboot. What should I do to make the /dev/dsk permanent for the snapshot disk?

Larry

[Steven E. Protter]

Shalom Larry,

Either there is an error in the SAN configuration or there are two paths to the disk (alternate links) in a way I've never seen before.

ioscan -fnC disk

I'm betting on a SAN configuration error.

Alternate: driver problem with the hbc(fiber channel interface).

SEP

[Larry J. Griswold]

We do have two fibre channel adapters, two fibre switches and two HSV100 controllers, so we do see with ioscan -f four paths to all our SAN disk. The LUN always is presented the same

# ioscan -f
.
.
hsx 23 0/6/1/0.10.0.1.0.0.6 hsx CLAIMED DEVICE HP HSV100

but the ioscan -fnC disk shows different /dev/dsk's

# ioscan -fnC disk
.
.
disk 10 255/255/0/0.7 sdisk CLAIMED DEVICE HSV100 HP

# ll /dev/dsk
total 0
brw-r----- 1 bin sys 31 0x000000 Aug 15 2005 c0t0d0
brw-r----- 1 bin sys 31 0x0a0000 Nov 14 11:23 c10t0d0
brw-r----- 1 bin sys 31 0x0a0100 Nov 14 11:23 c10t0d1
brw-r----- 1 bin sys 31 0x0a0200 Nov 14 11:23 c10t0d2
brw-r----- 1 bin sys 31 0x0a0300 Nov 14 11:23 c10t0d3
brw-r----- 1 bin sys 31 0x0a0400 Jan 18 15:01 c10t0d4
brw-r----- 1 bin sys 31 0x0a0500 Apr 6 10:06 c10t0d5
brw-r----- 1 bin sys 31 0x0a0600 Apr 6 10:06 c10t0d6
brw-r----- 1 bin sys 31 0x0a0700 May 10 14:30 c10t0d7
brw-r----- 1 bin sys 31 0x0a1000 May 10 14:30 c10t1d0
brw-r----- 1 bin sys 31 0x020000 Aug 15 2005 c2t0d0
brw-r----- 1 bin sys 31 0x021000 Aug 15 2005 c2t1d0

notice that the c10t0d7 was create yesterday

[BPatrick]

I suspect this should be a configuration problem of securepath. Which version of secure path do you use?

[m saravanan · Accepted solution]

Larry..

I hope, your setup doesn't have Secure Path Installed. It always suggested to access the LUNs through Secure Path. Since Secure Path Mapping is based on WWN number, the device file will be consistent across reboots.


Regds,
Saravanan

[Larry J. Griswold]

We are running SecurePath A.3.0E.00F.00F.

Maybe the WWID changes everytime we do a snapshot. Here is our process:

11:30pm HP-UX deletes the snapshot volume group.

11:35pm Storge Management Applicance runs SSSU to delete the snapshot and then recreate the snapshot

1:00am SSSU presents the new snapshot to the HP-UX system

1:15am HP-UX recreates the volumen group.

1:35am Data Protector starts the backup of the snapshot disk.


Is there a better way of doing this?

Larry

[BPatrick]

Have you made any changes in the fabric recently? or any configuration changes?
Also check for any latest patches for securepath.

[Larry J. Griswold]

We have been running on the same configuration since August of last year.

We have two rp4440 servers configured the same way and both have this same problem.

I will look into the latest securepath patches.

Larry

[Dave La Mar]

Larry -
We were hit by this after the original EVA installation.

Look at the following to make your configuration consistent acroos reboots -

# spmgr update
# kmadmin -L swspData
# config -M swspData -u

Best of luck.

Regards,

dl