One mirrored disk died, other not mounted at reboot

[Alan Solomon]

I have a pair of mirrored disks (non-boot) where one died. Ordinarily when this occurs, as I recall the remaining disk remains in use. This time, at least after reboot, the volume group (vg01) was "inactive", and to active it via vgchange I had to use the "-q n" option to ignore quorum enforcement, then I was able to mount the logical volume OK and limp along until I get a spare disk onsite. Is this normal? This system is on 11.0, prior experience with a failed mirrored drive was 10.20.

[A. Clay Stephenson · Accepted solution]

This is perfectly normal. With one disk out in a 2 disk volume group you don't have more than 50% of the disks available. All would have been well had you not powered-down or deactivated the volume group. You can, of course, force the activation as you did. If you are using hot-plug disks there was no reason to powerdown. I haven't shutdown to replace a disk (including boot disks) in well over five years.

[Jeff Schussele]

Hi Allan,

This is totally normal behavior.

If this was vg00 then on the mkboot -a command you'd use "hpux -lq" to boot up w/o quorum.

If this was an MC/SG system & this VG was not a cluster VG then you could use the string
/sbin/vgchange -a y -q n /dev/vg_name
in the /etc/lvmrc file to activate this VG w/o quorum.
I'm not sure how to do this without MC/SG in play. I think it's deliberate to alert you to a failed disk in case you're not using EMS or some other monitoring/alerting SW.

Rgds,
Jeff

[Alan Solomon]

Thanks for the reassurances! I guess I never rebooted when a mirrored disk failed in the past. Actually this system is in Shenzhen, China and I'm in Michigan. A Chinese colleague performed the reboot then called me.

[Todd McDaniel_1]

Alan,

Just a comment or two.

If you had a disk available AND This is a recent model system (K,L,N,V, etc...) on 11.0, you could have replaced the failed disk while online. OR you could have merely shutdown -h instead of -ry to add the disk in while the box was down.

I'm only guessing that you rebooted to get the box back in service ASAP.

BTW, a trick I use is to build my VGs (ie VG01, VGO2, etc) with more than one LVOL (/home, /tmp, /crash, etc) so that the VG always has more than 2 disks in it so you dont have this problem in the future.