[linux-lvm] LVM don't work anymore (wrong version?)

[linux-lvm] LVM don't work anymore (wrong version?)

[Jean-Eric Cuendet]

Hi,
I'm using LVM 0.9.1_beta3 since 2 weeks now without any problem.
I'm having a problem now.
I have one VG and one LV in it.
I need to create another VG made of 3 PVs.
I issue:
[root@fatboy hosts]# pvcreate -f /dev/md7
pvcreate -- ERROR: VGDA in kernel and lvmtab are NOT consistent; please run
vgscan
What's that?

When running vgscan:
[root@fatboy hosts]# vgscan
vgscan -- reading all physical volumes (this may take a while...)
vgscan -- no volume groups found

What can I do with that?
Thanks
-jec

PS: I already tried to upgrade to beta4 (only the tools). If you want me to
upgrade the kernel to beta4, wait this evening :-) (prod machine)

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
Jean-Eric Cuendet
Linkvest SA
Av des Baumettes 19, 1020 Renens Switzerland
Tel +41 21 632 9043  Fax +41 21 632 9090
http://www.linkvest.com  E-mail: jean-eric.cuendet@linkvest.com
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

RE: [linux-lvm] LVM don't work anymore (wrong version?)

[Jean-Eric Cuendet]

> > pvcreate -- ERROR: VGDA in kernel and lvmtab are NOT 
> consistent; please run
> > vgscan
> > What's that?
> 
> You shouldn't be able to get that message with LVM 0.9.1 beta3 at all.
> Are you by accident using an older version?

No, I'm sure it's beta3 and after beta4 (I tried it, just in case)

> If vgscan still complains and you have your 
> /etc/lvmconf/vg.conf (assuming
> that your VG is named vg :) VGDA backup file you can reactivate your
> VG with:
> 
> rm -rf /etc/lvmtab*
> mkdir /etc/lvmtab.d
> echo -ne "vg\0" > /etc/lvmtab
> cp /etc/lvmconf/vg.conf /etc/lvmtab.d/vg
> vgchange -ay

It works! Thanks.
The problem is that vgscan creates a file /etc/lvmtab without the VG name.
The file had only \0 in it!
I think that there is a bug. Would you like that I try somethng else?
I'll reboot the server this evening, so I'll have beta4 for tools AND
kernel.
Thanks for your help.
-jec

Re: [linux-lvm] LVM don't work anymore (wrong version?)

[Heinz J. Mauelshagen]

On Wed, Feb 14, 2001 at 11:15:53AM +0100, Jean-Eric Cuendet wrote:
> 
> Hi,
> I'm using LVM 0.9.1_beta3 since 2 weeks now without any problem.
> I'm having a problem now.
> I have one VG and one LV in it.
> I need to create another VG made of 3 PVs.
> I issue:
> [root@fatboy hosts]# pvcreate -f /dev/md7
> pvcreate -- ERROR: VGDA in kernel and lvmtab are NOT consistent; please run
> vgscan
> What's that?

You shouldn't be able to get that message with LVM 0.9.1 beta3 at all.
Are you by accident using an older version?

> 
> When running vgscan:
> [root@fatboy hosts]# vgscan
> vgscan -- reading all physical volumes (this may take a while...)
> vgscan -- no volume groups found
> 
> What can I do with that?
> Thanks
> -jec
> 
> PS: I already tried to upgrade to beta4 (only the tools). If you want me to
> upgrade the kernel to beta4, wait this evening :-) (prod machine)

Yes, please try that.

If vgscan still complains and you have your /etc/lvmconf/vg.conf (assuming
that your VG is named vg :) VGDA backup file you can reactivate your
VG with:

rm -rf /etc/lvmtab*
mkdir /etc/lvmtab.d
echo -ne "vg\0" > /etc/lvmtab
cp /etc/lvmconf/vg.conf /etc/lvmtab.d/vg
vgchange -ay

> 
> _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
> Jean-Eric Cuendet
> Linkvest SA
> Av des Baumettes 19, 1020 Renens Switzerland
> Tel +41 21 632 9043  Fax +41 21 632 9090
> http://www.linkvest.com  E-mail: jean-eric.cuendet@linkvest.com
> _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
> 
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] LVM don't work anymore (wrong version?)

[Heinz J. Mauelshagen]

On Wed, Feb 14, 2001 at 05:45:04PM +0100, Jean-Eric Cuendet wrote:
> 
> > > pvcreate -- ERROR: VGDA in kernel and lvmtab are NOT 
> > consistent; please run
> > > vgscan
> > > What's that?
> > 
> > You shouldn't be able to get that message with LVM 0.9.1 beta3 at all.
> > Are you by accident using an older version?
> 
> No, I'm sure it's beta3 and after beta4 (I tried it, just in case)
> 
> > If vgscan still complains and you have your 
> > /etc/lvmconf/vg.conf (assuming
> > that your VG is named vg :) VGDA backup file you can reactivate your
> > VG with:
> > 
> > rm -rf /etc/lvmtab*
> > mkdir /etc/lvmtab.d
> > echo -ne "vg\0" > /etc/lvmtab
> > cp /etc/lvmconf/vg.conf /etc/lvmtab.d/vg
> > vgchange -ay
> 
> It works! Thanks.

You're welcome.

> The problem is that vgscan creates a file /etc/lvmtab without the VG name.
> The file had only \0 in it!
> I think that there is a bug. Would you like that I try somethng else?

The problem you are facing is probably related to a PV UUID bug we
recently fixed.
Send me the output of:

pvscan
pvscan -u

before changing anything in your VG.

> I'll reboot the server this evening, so I'll have beta4 for tools AND
> kernel.
> Thanks for your help.
> -jec
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

RE: [linux-lvm] LVM don't work anymore (wrong version?)

[Jean-Eric Cuendet]

Hi,
I have this same problem on two of our servers! The 2 runs the same kernel:
2.4.1 with ACL patch and LVM 0.9.1beta4

With the second one, I can do much more tests than the other. It's a
semi-prod machine.
Here is what I did:

[root@zorro /root]# pvscan
pvscan -- reading all physical volumes (this may take a while...)
pvscan -- ACTIVE   PV "/dev/md6"  is associated to an unknown VG (run
vgscan)
pvscan -- ACTIVE   PV "/dev/md7"  is associated to an unknown VG (run
vgscan)
pvscan -- ACTIVE   PV "/dev/md8"  is associated to an unknown VG (run
vgscan)
pvscan -- total: 3 [8.8 GB] / in use: 3 [8.8 GB] / in no VG: 0 [0]
 
[root@zorro /root]# pvscan -u
pvscan -- reading all physical volumes (this may take a while...)
pvscan -- ACTIVE   PV "/dev/md6" with UUID
"v7Y8gz-KMXB-d6Iq-Y0U1-iiSo-javD-wOL89P"  is associated to an
unknown VG (run vgscan)
pvscan -- ACTIVE   PV "/dev/md7" with UUID
"xTl7Nc-6DhQ-KV3c-HHd0-LI50-nQwq-wFBYCG"  is associated to an
unknown VG (run vgscan)
pvscan -- ACTIVE   PV "/dev/md8" with UUID
"oDXPq6-s9Zn-dOHF-ss8J-Rdxt-jTKT-wFoIGQ"  is associated to an
unknown VG (run vgscan)
pvscan -- total: 3 [8.8 GB] / in use: 3 [8.8 GB] / in no VG: 0 [0]
 
[root@zorro /root]# vgscan
vgscan -- reading all physical volumes (this may take a while...)
vgscan -- no volume groups found
 
[root@zorro /root]# vgcreate
vgcreate -- ERROR: VGDA in kernel and lvmtab are NOT consistent; please run
vgscan
 
[root@zorro /root]#


What to do now?  I wait for your help!
Thanks
-jec

Re: [linux-lvm] LVM don't work anymore (wrong version?)

[Andreas Dilger]

Jean-Eric Cuendet writes:
> I have this same problem on two of our servers! The 2 runs the same kernel:
> 2.4.1 with ACL patch and LVM 0.9.1beta4
> 
> [root@zorro /root]# pvscan -u
> pvscan -- reading all physical volumes (this may take a while...)
> pvscan -- ACTIVE   PV "/dev/md6" with UUID
> "v7Y8gz-KMXB-d6Iq-Y0U1-iiSo-javD-wOL89P"  is associated to an
> unknown VG (run vgscan)
> pvscan -- ACTIVE   PV "/dev/md7" with UUID
> "xTl7Nc-6DhQ-KV3c-HHd0-LI50-nQwq-wFBYCG"  is associated to an
> unknown VG (run vgscan)
> pvscan -- ACTIVE   PV "/dev/md8" with UUID
> "oDXPq6-s9Zn-dOHF-ss8J-Rdxt-jTKT-wFoIGQ"  is associated to an
> unknown VG (run vgscan)

Try running 'pvdata -U' on each of the PV's, and see if there are empty
UUID slots on some of them.  This is a bug with the old LVM tools.  If
this is true, you can try to comment out the section in pv_read_all_pv_of_vg
which verifies the UUIDs (starts like "if (uuids) {"), and try vgscan
again.

Cheers, Andreas
-- 
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert

RE: [linux-lvm] LVM don't work anymore (wrong version?)

[Jean-Eric Cuendet]

> Try running 'pvdata -U' on each of the PV's, and see if there 
> are empty
> UUID slots on some of them.  This is a bug with the old LVM tools.  If

I have 13 slots and the 13th is EMPTY. Is that what you mean?

> this is true, you can try to comment out the section in 
> pv_read_all_pv_of_vg
> which verifies the UUIDs (starts like "if (uuids) {"), and try vgscan
> again.

I did it and vgscan output nothing about Volume Groups!
Output of debug vgscan is attached.

Here is the output of a vgchange:
[root@fatboy 0.9.1_beta4]# echo -ne "NOTSURE_HDX\0" > /etc/lvmtab
[root@fatboy 0.9.1_beta4]# vgchange -ay
vgchange -- ERROR: physical volume "/dev/hdc9" of volume group "NOTSURE_HDX"
is inconsistent
vgchange -- run vgscan

FYI, I have hda1, hda5->hda10 + hdc1, hdc5->hdc10 in the VG NOTSURE_HDX.
hdc9 seems broken, but why? And how to restore?

What do you think?
I'm in big trouble no?

Bye
-jec

Re: [linux-lvm] LVM don't work anymore (wrong version?)

[Andreas Dilger]

Jean-Eric writes:
> > Try running 'pvdata -U' on each of the PV's, and see if there are empty
> > UUID slots on some of them.  This is a bug with the old LVM tools.  If
> 
> I have 13 slots and the 13th is EMPTY. Is that what you mean?

Yes.  This was a bug in the old (0.9.1beta2?) LVM user tools.  The problem is
that this data was written to disk incorrectly, and now vgscan doesn't work.

> > this is true, you can try to comment out the section in 
> > pv_read_all_pv_of_vg which verifies the UUIDs (starts like
> > "if (uuids) {"), and try vgscan again.
> 
> I did it and vgscan output nothing about Volume Groups!
> Output of debug vgscan is attached.

I don't see any attachment.  I was looking at the pv_read_all_pv_of_vg()
code, and the section actually starts "if (uuids > 0) {", so you can
replace this with "if (0) {" and the UUID checking will be skipped.
This _should_ fix your problem, but it is possible there is also another
problem as well.

> Here is the output of a vgchange:
> [root@fatboy 0.9.1_beta4]# echo -ne "NOTSURE_HDX\0" > /etc/lvmtab
> [root@fatboy 0.9.1_beta4]# vgchange -ay
> vgchange -- ERROR: physical volume "/dev/hdc9" of volume group "NOTSURE_HDX"
> is inconsistent
> vgchange -- run vgscan
> 
> FYI, I have hda1, hda5->hda10 + hdc1, hdc5->hdc10 in the VG NOTSURE_HDX.
> hdc9 seems broken, but why? And how to restore?

You can try restoring your VGDA with "vgcfgrestore /dev/hdc9".  You may
have to restore the VGDA on all of the disks at the same time, I'm not
sure.  The good news is that if you do a vgcfgrestore with the new (beta4)
tools it will also fix the UUID problem above.

Cheers, Andreas

PS - it is strange that you have multiple partitions on the same disk as
     separate PVs.  It is the most efficient (space wise) to only make a
     single large PV on each disk.
-- 
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert