Failure to attach PV to the root volume group on boot

[EEESEC]

During boot, I am getting the below message. No matter what I do, it keeps initially looking for the device with minor number 0x1000010. It finds the correct one after a scan (0x100001e) but this will not persist. At next reboot, it looks for x1000010 again. This is a boot from SAN configuration (EVA 4400). Since it still boots this in not critical, but is annoying.

I also have duplicate DSFs for some reason as well. When I try to rmsf any of them it states they are busy. I guess I could try from lvm maintence mode.... As far as I know there is no way to change a minor number for a persistent DSF, yes? Any advice would be appreciated.

SECLLDV01-root-(/home/root)-$ll /dev/disk/
total 0
brw-r----- 1 bin sys 1 0x00001d Apr 7 13:53 disk0
brw-r----- 1 bin sys 1 0x00001f Apr 7 14:30 disk0_p1
brw-r----- 1 bin sys 1 0x00001e Apr 7 14:30 disk0_p2
brw-r----- 1 bin sys 1 0x000020 Apr 7 14:30 disk0_p3
brw-r----- 1 bin sys 1 0x00001d Apr 4 14:57 disk15
brw-r----- 1 bin sys 1 0x00001f Apr 4 14:57 disk15_p1
brw-r----- 1 bin sys 1 0x00001e Apr 4 14:57 disk15_p2
brw-r----- 1 bin sys 1 0x000020 Apr 4 14:57 disk15_p3


vmunix: LVM : Failure in attaching PV (dev=0x1000010) to the root volume group.
The physical volume does not exist, or is not configured in the kernel
vmunix: LVM : Activation of root volume group failed
vmunix: LVM: Scanning for Root VG PVs (VGID 0x27ed8ed0 0x4cbf5813)
vmunix: LVM: Rootvgscan detected 1 PV(s). Will attempt root VG activation using
vmunix: the following PV(s):
vmunix: 0x100001e
vmunix: LVM: Root VG activated

[Viktor Balogh]

what is your patch level?
is PHCO_27369 installed?

ftp://us-ffs.external.hp.com/patch_bundles/hp-ux/HWE/HWE1100_B.11.00.0403.3.txt

it describes a similar problem...

[EEESEC]

I should have mentioned this is 11iv3. The below patch wouldn't apply, but thanks for the tip.

I fixed this by booting into lvm maintenance mode and manually removed the duplicate dsfs (had to use rm -R and not rmsf). The I did a vgscan -Nf to update lvmtab entries. (However this did not work initially, it kept telling me I had to use the -N option even though I was. This is actually a bug with lvm, updating to the most current version of lvm fixed this).

Then I rebooted and all seems well.

[Michael Steele_2]

Sound like /etc/lvmtab is old. Use strings /etc/lvmtab to review the luns and vg's currently expected to be seen by LVM.

Use vgscan -apv to 'preview' a new lvmtab.

In short, someone disconnected some luns and this is common in a SAN environment.

[EEESEC]

Yes, as mentioned above I had to run the vgscan (once I got it working) to update the lvmtab with the proper dsf.

Closed.