Volume Groups won't stay active

[Dan Bolton]

I recently upgraded one of our K370 servers from HPUX 10.20 to 11.00 (cold install). Of course the previous non-root volume groups no longer existed, so I performed the following series of commands to restore each of them:

mkdir /dev/vg##
mknod /dev/vg##/group c 64 0x??0000
vi mapfile - Edit the mapfile with logical volume names
vgimport -m mapfile -v /dev/vg## /dev/dsk/cXtXdX
vgchange -a y /dev/vg##
mkdir /mountpoint
mount -F vxfs /dev/vg##/lv_name /mountpoint
vgcfgbackup /dev/vg##
vi /etc/fstab - Edit fstab file to automount at reboot

The problem is, that every time I reboot the system I get the following error:

mountall: cannot mount /dev/vg05/lvol01
mountall: diagnostics from mount
mount - Cannot open /dev/vg05/lvol01: No such device or address

When I check, the vg's exist but are not "active". I can run the vgchange command to activate them, and use mount -a and they mount fine (after fsck).

I have used a similar process to upgrade another K370 and the disks mount fine on reboot. Did I miss something on this one?

Attached is /etc/fstab showing the vg's involved. The vg HDS_LU2.R0 is a FibreChannel attached disk array, vg03 and vg05 are SCSI attached local disks.

Thanks,
Dan

[Tom Danzig]

Your logical volume names do nat match. Per your fstab:

/dev/vg05/lvol_mars /u10 vxfs rw,suid,delaylog,datainlog 0 2
/dev/vg05/lvol_oracle /oracle vxfs rw,suid,delaylog,datainlog 0 2
/dev/vg05/lvol_baan /B4c3 vxfs rw,suid,delaylog,datainlog 0 2
/dev/vg05/lvol_baanDD /B4c3DD vxfs rw,suid,delaylog,datainlog 0 2
/dev/vg05/lvol_rmc /rmc vxfs rw,suid,delaylog,datainlog 0 2


Per the mount message:
Cannot open /dev/vg05/lvol01

Something went afoul during the import process using the mapfile.

If you know which LV's are which based on minor number, you could do a mv on the lvol0# files in /dev/vg05 or modify the fstab to match.

[Dan Bolton]

Point of Clarification:

The error message listed is only a representation. I receive an error for each lv on a non-root vg.

Thanks,
Dan

[James A. Donovan]

Check your /etc/lvmrc file. Make sure that AUTO_VG_ACTIVATE is set to 1. If it's 0, then only vg00 will be automatically actived at boot time.

[Dan Bolton]

Jim,

I checked /etc/lvmrc and AUTO_VG_ACTIVATE = 1

I also checked the permissions on /etc/lvmrc to make sure it was readable.

Dan

[James A. Donovan]

hmmm...after a reboot, try running the /sbin/lvmrc script manually and see what happens.

Finally, try rebuilding the /etc/lvmtab file using vgscan.

# mv /etc/lvmtab /etc/lvmtab.old
# vgscan -av

[Patrick Wallek]

Check /var/adm/syslog/syslog.log, dmesg and /etc/rc.log and see if there are any errors recorded when it attempts to activate the VGs. If there are, please post them.

[Dan Bolton]

Thanks for the ideas, I will try to get some time to work on this during the holidays, but may not be able to until after the 1st. I will post updates when I have something.

'til then... Happy Holidays to All!!

db