[linux-lvm] HELP: vgscan segfaults

[linux-lvm] HELP: vgscan segfaults

[Peter Lohmann]

Hi folks!

I cannot access my lv's anymore, because vgscan segfaults.
I used Knoppix (2.4.20 Kernel) yesterday - it worked fine.
But today, when I try to do a vgscan -v i get the following output:

root@tty1[/]# vgscan -v
vgscan -- removing "/etc/lvmtab" and "/etc/lvmtab.d"
vgscan -- creating empty "/etc/lvmtab" and "/etc/lvmtab.d" 
vgscan -- reading all physical volumes (this may take a while...)
vgscan -- scanning for all active volume group(s) first
vgscan -- reading data of volume group "myvg" from physical volume(s)
Segmentation fault

As i said, yesterday it worked fine?!?

I'm using debian packages lvm-common 1.5.5 and lvm10 v. 1.0.4-6

Thanks!

Peter

Re: [linux-lvm] HELP: vgscan segfaults

[Patrick Caulfield]

On Wed, Jan 29, 2003 at 04:12:09PM +0100, Peter Lohmann wrote:
> Hi folks!
> 
> I cannot access my lv's anymore, because vgscan segfaults.
> I used Knoppix (2.4.20 Kernel) yesterday - it worked fine.
> But today, when I try to do a vgscan -v i get the following output:
> 
> root@tty1[/]# vgscan -v
> vgscan -- removing "/etc/lvmtab" and "/etc/lvmtab.d"
> vgscan -- creating empty "/etc/lvmtab" and "/etc/lvmtab.d" 
> vgscan -- reading all physical volumes (this may take a while...)
> vgscan -- scanning for all active volume group(s) first
> vgscan -- reading data of volume group "myvg" from physical volume(s)
> Segmentation fault
> 
> As i said, yesterday it worked fine?!?
> 
> I'm using debian packages lvm-common 1.5.5 and lvm10 v. 1.0.4-6

Are there any kernel messages (dmesg) ?

patrick

Re: [linux-lvm] HELP: vgscan segfaults

[Peter Lohmann]

On Wed, Jan 29, 2003 at 03:18:19PM +0000, Patrick Caulfield wrote:
> On Wed, Jan 29, 2003 at 04:12:09PM +0100, Peter Lohmann wrote:
> > Hi folks!
> > 
> > I cannot access my lv's anymore, because vgscan segfaults.
> > I used Knoppix (2.4.20 Kernel) yesterday - it worked fine.
> > But today, when I try to do a vgscan -v i get the following output:
> > 
> > root@tty1[/]# vgscan -v
> > vgscan -- removing "/etc/lvmtab" and "/etc/lvmtab.d"
> > vgscan -- creating empty "/etc/lvmtab" and "/etc/lvmtab.d" 
> > vgscan -- reading all physical volumes (this may take a while...)
> > vgscan -- scanning for all active volume group(s) first
> > vgscan -- reading data of volume group "myvg" from physical volume(s)
> > Segmentation fault
> > 
> > As i said, yesterday it worked fine?!?
> > 
> > I'm using debian packages lvm-common 1.5.5 and lvm10 v. 1.0.4-6
> 
> Are there any kernel messages (dmesg) ?

No :-(

Peter

Re: [linux-lvm] HELP: vgscan segfaults

[Patrick Caulfield]

On Wed, Jan 29, 2003 at 04:32:29PM +0100, Peter Lohmann wrote:
> On Wed, Jan 29, 2003 at 03:18:19PM +0000, Patrick Caulfield wrote:
> > On Wed, Jan 29, 2003 at 04:12:09PM +0100, Peter Lohmann wrote:
> > > Hi folks!
> > > 
> > > I cannot access my lv's anymore, because vgscan segfaults.
> > > I used Knoppix (2.4.20 Kernel) yesterday - it worked fine.
> > > But today, when I try to do a vgscan -v i get the following output:
> > > 
> > > root@tty1[/]# vgscan -v
> > > vgscan -- removing "/etc/lvmtab" and "/etc/lvmtab.d"
> > > vgscan -- creating empty "/etc/lvmtab" and "/etc/lvmtab.d" 
> > > vgscan -- reading all physical volumes (this may take a while...)
> > > vgscan -- scanning for all active volume group(s) first
> > > vgscan -- reading data of volume group "myvg" from physical volume(s)
> > > Segmentation fault
> > > 
> > > As i said, yesterday it worked fine?!?
> > > 
> > > I'm using debian packages lvm-common 1.5.5 and lvm10 v. 1.0.4-6
> > 
> > Are there any kernel messages (dmesg) ?
> 
> No :-(

If your metadata backups are up-to-date then "vgchange -ay" should activate the
volumes regardless. Something must have changed, programs don't just work one
day and segv the next without good reason.

-- 

patrick

Re: [linux-lvm] HELP: vgscan segfaults

[Peter Lohmann]

On Wed, Jan 29, 2003 at 03:42:07PM +0000, Patrick Caulfield wrote:
> On Wed, Jan 29, 2003 at 04:32:29PM +0100, Peter Lohmann wrote:
> > On Wed, Jan 29, 2003 at 03:18:19PM +0000, Patrick Caulfield wrote:
> > > On Wed, Jan 29, 2003 at 04:12:09PM +0100, Peter Lohmann wrote:
> > > > Hi folks!
> > > > 
> > > > I cannot access my lv's anymore, because vgscan segfaults.
> > > > I used Knoppix (2.4.20 Kernel) yesterday - it worked fine.
> > > > But today, when I try to do a vgscan -v i get the following output:
> > > > 
> > > > root@tty1[/]# vgscan -v
> > > > vgscan -- removing "/etc/lvmtab" and "/etc/lvmtab.d"
> > > > vgscan -- creating empty "/etc/lvmtab" and "/etc/lvmtab.d" 
> > > > vgscan -- reading all physical volumes (this may take a while...)
> > > > vgscan -- scanning for all active volume group(s) first
> > > > vgscan -- reading data of volume group "myvg" from physical volume(s)
> > > > Segmentation fault
> > > > 
> > > > As i said, yesterday it worked fine?!?
> > > > 
> > > > I'm using debian packages lvm-common 1.5.5 and lvm10 v. 1.0.4-6
> > > 
> > > Are there any kernel messages (dmesg) ?
> > 
> > No :-(
> 
> If your metadata backups are up-to-date then "vgchange -ay" should activate the
> volumes regardless. Something must have changed, programs don't just work one
> day and segv the next without good reason.

I know... but still:

--> vgchange -ay myvg
vgchange -- volume group "myvg" does not exist

Same with vgchange -ay

:-(

Re: [linux-lvm] HELP: vgscan segfaults

[Patrick Caulfield]

On Wed, Jan 29, 2003 at 04:59:36PM +0100, Peter Lohmann wrote:
> > 
> > If your metadata backups are up-to-date then "vgchange -ay" should activate the
> > volumes regardless. Something must have changed, programs don't just work one
> > day and segv the next without good reason.
> 
> I know... but still:
> 
> --> vgchange -ay myvg
> vgchange -- volume group "myvg" does not exist
> 
> Same with vgchange -ay
> 
> :-(

Looks like the segfaulting vgscan started to write the metadata backups. See of
you can find valid ones in /etc/lvmtab and restore them using vgcfgrestore.

OR it might be as easy as creating a new /etc/lvmtab with "myvg\0" in it and
retrying vgchange.
-- 

patrick

Re: [linux-lvm] HELP: vgscan segfaults

[Heinz J . Mauelshagen]

On Wed, Jan 29, 2003 at 04:09:58PM +0000, Patrick Caulfield wrote:
> On Wed, Jan 29, 2003 at 04:59:36PM +0100, Peter Lohmann wrote:
> > > 
> > > If your metadata backups are up-to-date then "vgchange -ay" should activate the
> > > volumes regardless. Something must have changed, programs don't just work one
> > > day and segv the next without good reason.
> > 
> > I know... but still:
> > 
> > --> vgchange -ay myvg
> > vgchange -- volume group "myvg" does not exist
> > 
> > Same with vgchange -ay
> > 
> > :-(
> 
> Looks like the segfaulting vgscan started to write the metadata backups. See of
> you can find valid ones in /etc/lvmtab and restore them using vgcfgrestore.
> 
> OR it might be as easy as creating a new /etc/lvmtab with "myvg\0" in it and
> retrying vgchange.

...plus "mkdir /etc/lvmtab.d;cp /etc/lvmconf/myvg.conf /etc/lvmtab.d/myvg"
before the "vgchange -ay".

Additionally you need to change the LVM config (i.e. by running
"lvchange -pr .." or something) so that the ondisk metadata gets written
back to the PVs.

> -- 
> 
> patrick
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

-- 

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

[linux-lvm] lvm and fai

[Sébastien GALLET]

Hello guys,
This message is for the debian maintainer of lvm package.
I've got a little problem using fai (Fully Automatic Installation) and lvm.
My server runs lvm.
When creating fai-nfsroot (using debootstrap) with lvm activated, 
debootstrap stops with this message :

# vgcfgbackup -- ERROR: VGDA in kernel and lvmtab are NOT consistent; 
please run vgscan
# dpkg: error processing lvm10 (--configure):
# subprocess post-installation script returned error exit status 98
# Errors were encountered while processing:
# lvm10

I suppose that my server configuration interferes with the one in 
debootstrap.

Why include vgcfgbackup in postinst scripts ? Is it possible do it in 
"silent" mode even if there's errors ?

Sorry for speaking so badly English :):)

TIA

Re: [linux-lvm] lvm and fai

[Patrick Caulfield]

On Tue, Feb 04, 2003 at 03:06:46PM +0100, S?bastien GALLET wrote:
> Hello guys,
> This message is for the debian maintainer of lvm package.
> I've got a little problem using fai (Fully Automatic Installation) and lvm.
> My server runs lvm.
> When creating fai-nfsroot (using debootstrap) with lvm activated, 
> debootstrap stops with this message :
> 
> # vgcfgbackup -- ERROR: VGDA in kernel and lvmtab are NOT consistent; 
> please run vgscan
> # dpkg: error processing lvm10 (--configure):
> # subprocess post-installation script returned error exit status 98
> # Errors were encountered while processing:
> # lvm10
> 
> I suppose that my server configuration interferes with the one in 
> debootstrap.
> 
> Why include vgcfgbackup in postinst scripts ? Is it possible do it in 
> "silent" mode even if there's errors ?
> 

The reason that vgcfgbackup is run is to make upgrades as painless as
possible by ensuring that people have the latest metadata backups available
(you'd be surprised how many people have been hit by that).

I'm not familiar with fai so I'm not sure what is actually happening here. I
suppose the metadata backups in the chroot are different to what the host
machine is really using but I'm not sure how they got there in the first place
or what they are based on. Maybe you could copy the server's metadata backups
into the chroot to get around the problem (it shouldn't interfere with the
target system as it will do a vgscan when it first boots up).

Another thing you could do is build a customised package without the
vgcfgbackup in postinst (that's easier than it sounds, simply apt-get source
lvm10, edit lvm10-<version>/debian/lvm.postinst, then dpkg-buildpackage
-rfakeroot)

-- 

patrick

Re: [linux-lvm] HELP: vgscan segfaults

[Terje Kvernes]

listenmann@elrond.lohmann-services.de (Peter Lohmann) writes:

> root@tty1[/]# vgscan -v
> vgscan -- removing "/etc/lvmtab" and "/etc/lvmtab.d"
> vgscan -- creating empty "/etc/lvmtab" and "/etc/lvmtab.d" 
> vgscan -- reading all physical volumes (this may take a while...)
> vgscan -- scanning for all active volume group(s) first
> vgscan -- reading data of volume group "myvg" from physical volume(s)
> Segmentation fault

  I've seen this recently as well.  on my end there were some issues
  with a PV getting lost.  the problem in my end was that the
  partition table of a drive was lost somehow, and without a partition
  table you don't get any PVs.  :-)

  I recreated the partitontable (fdisk), recreated the PV on it
  (pvcreate) and added the PV to the appropriate VG (vgextend).  this
  happily enough didn't touch the data on the drive, so the LVs that
  resided on the readded drive where found and worked as normal
  afterwards.  

-- 
Terje