volume group ghost disk hp-ux 10.20

[Brian Harris]

I am haveing problems with a volume group after a mirrored disk failure. The volume group is vg00. The Curr PV 5 ACT PV 4. Their are only 4 disk in system right now. The lvmtab currently shows just the 4 good disk.
I have already preformed the vgreduce and vgscan commands, but all other commands when execute say error similar to the following:
vgdisplay: Warning: couldn't query physical volume "":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.

When I try to a vgcfgbackup it fails and gives
similar message as above, the same lvlnboot -v
command. I also can no longer create a logical volume or file system on that volume group

I had replace bad disk and tried to restore, but that failed, not sure why, but at this time I have removed that disk and I am just trying to get system to work correctly without that disk and then I will re do mirroring of disk from scratch.
The vg00.conf that I have is a good one from before problem happened, but have not been able to remake it with that disk removed.

As an additional note during intial try at fixing it I was have problems during actual re mirroring a lvol, I ended up removing that lvol and I have not been able to recreate it since then. This lvol is not part of hp-ux root, but had all of the customers executable codes. So I created a link from that location to another location where I restored data so that they could work. That logical volume does not show up in vgdisplay, but I presume it is still in vg00.conf

[Steven E. Protter]

Shalom,

cd /etc
mv lvmtab lvmtab.old
vgscan -a

This normally fixes it.

HP-UX 10.20 with no support for almost 5 years. Gutsy play.

SEP

[Brian Harris]

As stated earlier I have already done that, and the lvmtab seems to be correct, but the vg00.con is not correct and vgcfgbackup does not create a new vg00.conf, it gives message similar to one list from vgdisplay and does not say why it did not recreate vg00.conf.
As far as hp-ux 10.2 I understand, But Customer does not want to spend the money needed to convert application software

[Pupil_1]

try
vgreduce -f vg00

and if that does not fix it, try
vgreduce -l vg00 .

From vgreduce 11.23 Manpage.

-l pv_path This option removes the specified pv_path/s from the lvmtab file. This task will only be performed if the pv_path is currently marked as missing from the volume group. This option was mainly designed for the problem where the vgscan or the vgimport command place too many links beyond the max limit allowed in the lvmtab file. Currently the max limit is 8 paths to a PV (seven alternates and one primary). In this situation invoking the command without the -f option will not resolve the condition because the path is not attached to the volume group. Similarly the condition would not be overcome by invoking with the -f option as this works on PVs rather than links.



Not sure if this option is available with 10.2, I don't have to 10.2 to check.

Cheers.

[Brian Harris]

As stated earlier I have run vgreduce and I did run it with -f option. I believe this made the lvmtab correct, but vg00.conf does not list that disk but CUR and ACT PV values do not agree.
10.20 does not have the -l option.

[OldSchool]

Brian,

Take a look at this doc, particularly the part about reducing w/ the key number. I believe this will work w/ 10.20, but can't verify
http://www11.itrc.hp.com/service/cki/docDisplay.do?docLocale=en&docId=emr_na-c01182373-2

[Brian Harris]

I can't see document specified, link says you need a current support agrrement with HP,
I do not have one anymore. But I am not sure what there is to reduce. I did a vgreduce -f and when I recreated lvmtab it no longer had bad disk in it. Warning messages are now saying disk is called "" and
when you do strings on vg00.conf it still shows bad disk, but when do vgdisplay -v it does not.