vgcfgbackup: /etc/lvmtab is out of date with the running kernel

[John Kelly_3]

I have replaced and remirrored a boot disk in vg00 on an 11.00 system and am now getting the error: -
vgcfgbackup: /etc/lvmtab is out of date with the running kernel

I have run
vgreduce -f /dev/vg00
rm /etc/lvmtab
vgscan -v
and this hasn't rectified the situation. vgdisplay still shows
Cur PV 3
instead of 2 and the vgcfgbackup error still occurs.

What next?

[Peter Godron]

John,
this one seems to fit:
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=18493

[Eric Antunes]

Hi John,

Don't remove /etc/lvmtab: change its name instead:

mv /etc/lvmtab /etc/lvmtab.bak

And after that:

vgscan -av

And finally:

vgcfgbackup vg00


Best Regards,

Eric Antunes

[John Kelly_3]

Eric,
# mv /etc/lvmtab /etc/lvmtab.bak
# vgscan -av
Creating "/etc/lvmtab".
Couldn't stat physical volume "/dev/dsk/c3t2d0":
Invalid argument

/dev/vg00
/dev/dsk/c1t2d0
/dev/dsk/c2t2d0

/dev/vg02
/dev/dsk/c4t8d0

Scan of Physical Volumes Complete.
root_cycsuw:/ # vgcfgbackup vg00
vgcfgbackup: /etc/lvmtab is out of date with the running kernel:Kernel indicates 3 disks for "/dev/vg00"; /etc/lvmtab has 2 disks.
Cannot proceed with backup.

Peter, all I can see in that link that I haven't done already is the suggested vgexport/vgimport. That would have to be done in maintenance mode. I'll need to arrange some downtime.

[Eric Antunes]

Hi John,

You will need to vgextend this /dev/dsk/c3t2d0 before running vgscan -av.

See man vgextend and act accordingly.

Cheers,

Eric

[Court Campbell]

Try

vgchange -a y vg00

then retry vgscan.

[John Kelly_3]

Thanks for your replies Eric and Court but ...
Eric,
/dev/dsk/c3t2d0 is the DVD so I won't put that in vg00.
Court,
The volume group vg00 is already active.

I will try the export / import suggestion but need to wait for downtime.

[Court Campbell]

John I realize that vg00 is active. If you re-run the command it will update the kernel lvm structures. I thought you might give it a shot since your post states that the kernel and lvmtab are not consistent. Also, did your vgreduce -f actually work? I am not sure how you had your disks setup, but if you by chance had an lvol that spanned onto the disk that you replaced then vgreduce would spit out info about the lvol. You would more than likely need to recreate the lvol and restore what was there. Mind posting your lvmtab and vgdisplay -v vg00 output?

[John Kelly_3]

Court, Ok I understand why you asked me to try that but I had done it (vgchange -a y) previously. FYI the LVs were mirrored cleanly between the two disks.

In any case the server crashed and when rebooted the two OS disks were unusable. I had to restore from an ignite onto new disks.
So the condition was more than likely related to bad disks. :-( Which seems to be more and more common for me with these older servers.

Thanks for all your input. As always I have learned plenty from the experience. :-)

[Eric Antunes]

Hi John,

In case you are interessed, I have 10 36Gb disks and 4 9 Gb disks that were in 2 9000 L2000 servers.

Best Regards,

Eric

[John Kelly_3]

Thanks very much for the kind offer Eric, but there is no need. The machine is under a hardware support contract with a third-part company and it is their responsibility to provide the old 9G disks.

[John Kelly_3]

Disks replaced. Thanks for all the help.