Re: [linux-lvm] vgchange -a y /dev/System fails

Re: [linux-lvm] vgchange -a y /dev/System fails

[Marc MERLIN]

[Adding a Cc to lvm-devel, since I apparently hit a bug in the code. Please
remove or edit as appropriate]

On Mon, Jan 28, 2002 at 10:42:58AM +0100, Heinz J . Mauelshagen wrote:
> > It fails and says:
> > parameter error setting up snapshot copy on write exception
> > 
> > So yes, I have  snapshots in there, but I can't delete  the snapshots if the
> > volume isn't active, and I can't activate the volume because of a snapshot.
> > 
> > To make things worse, /dev/System contains /usr and /var.
> > 
> > The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17 with
> > 1.0.1rc4-ish
> 
> I am afraid you need to figure out the last metadata backup in /etc/lvmconf/
> which *doesn't* contain the flaky snapshot(s).
> 
> You didn't mention the name of you VG so let's say it was 'A'.
>
> You can do that with "vgcfgrestore -f /etc/lvmconf/A.conf.1.old -n A -ll".
> Replace the '1' in the metadata backup filename above with 2, 3 and so on
> (you should have a default of up to 9).
> After you found one (say #3), remember all you PVs belonging to that VG and run
> 
> pvcreate -yff /dev/AllYourPVsBelongingToA
> for pv in /dev/AllYourPVsBelongingToA
> do
>    vgcfgrestore -f /etc/lvmconf/A.conf.3.old -n A $pv
> done
> vgscan
> vgchange -ay A

Unfortunately, that doesn't work because I  had a cron job that was creating
daily snapshots (in an attempt to emulate ONTAP's snapshot's feature), so my
10 backups contain snapshot LVs

If the backups were text, I could edit them and remove the snapshot info in
there, but with a binary backup, I don't know how to do that.

If I  know how big  my /var and  /usr partitions were  and in which  order I
created them (vgcfgrestore -ll shows this  anyway), can I recreate the VG on
top of itself without losing the data in the the LVs?
Is there another way to recover from this?

More importantly, is this a known bug? Can I give info to help fix it?

Thanks,
Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

Re: [lvm-devel] Re: [linux-lvm] vgchange -a y /dev/System fails

[Marc MERLIN]

On Tue, Jan 29, 2002 at 02:01:25PM -0800, Marc MERLIN wrote:
> [Adding a Cc to lvm-devel, since I apparently hit a bug in the code. Please
> remove or edit as appropriate]

Really sorry about the dupe, I initially didn't post with my subscribed
address, and I thought the list was supposed to reject posts from a non
subscribed address.

Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

[linux-lvm] vgchange -a y /dev/System fails

[Marc MERLIN]

It fails and says:
parameter error setting up snapshot copy on write exception

So yes, I have  snapshots in there, but I can't delete  the snapshots if the
volume isn't active, and I can't activate the volume because of a snapshot.

To make things worse, /dev/System contains /usr and /var.

The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17 with
1.0.1rc4-ish

What's the magic override I need to use to make the volume active again?

Thanks
Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

Re: [linux-lvm] vgchange -a y /dev/System fails

[Heinz J . Mauelshagen]

Marc,

I am afraid you need to figure out the last metadata backup in /etc/lvmconf/
which *doesn't* contain the flaky snapshot(s).

You didn't mention the name of you VG so let's say it was 'A'.

You can do that with "vgcfgrestore -f /etc/lvmconf/A.conf.1.old -n A -ll".
Replace the '1' in the metadata backup filename above with 2, 3 and so on
(you should have a default of up to 9).
After you found one (say #3), remember all you PVs belonging to that VG and run

pvcreate -yff /dev/AllYourPVsBelongingToA
for pv in /dev/AllYourPVsBelongingToA
do
   vgcfgrestore -f /etc/lvmconf/A.conf.3.old -n A $pv
done
vgscan
vgchange -ay A

That should be it (crossing fingers ;-)


On Fri, Jan 25, 2002 at 03:35:33PM -0800, Marc MERLIN wrote:
> It fails and says:
> parameter error setting up snapshot copy on write exception
> 
> So yes, I have  snapshots in there, but I can't delete  the snapshots if the
> volume isn't active, and I can't activate the volume because of a snapshot.
> 
> To make things worse, /dev/System contains /usr and /var.
> 
> The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17 with
> 1.0.1rc4-ish
> 
> What's the magic override I need to use to make the volume active again?
> 
> Thanks
> Marc
> -- 
> Microsoft is to operating systems & security ....
>                                       .... what McDonalds is to gourmet cooking
>   
> Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 

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] vgchange -a y /dev/System fails

[Marc MERLIN]

[Adding a Cc to lvm-devel, since I apparently hit a bug in the code. Please
remove or edit as appropriate]

On Mon, Jan 28, 2002 at 10:42:58AM +0100, Heinz J . Mauelshagen wrote:
> > It fails and says:
> > parameter error setting up snapshot copy on write exception
> > 
> > So yes, I have  snapshots in there, but I can't delete  the snapshots if the
> > volume isn't active, and I can't activate the volume because of a snapshot.
> > 
> > To make things worse, /dev/System contains /usr and /var.
> > 
> > The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17 with
> > 1.0.1rc4-ish
> 
> I am afraid you need to figure out the last metadata backup in /etc/lvmconf/
> which *doesn't* contain the flaky snapshot(s).
> 
> You didn't mention the name of you VG so let's say it was 'A'.
>
> You can do that with "vgcfgrestore -f /etc/lvmconf/A.conf.1.old -n A -ll".
> Replace the '1' in the metadata backup filename above with 2, 3 and so on
> (you should have a default of up to 9).
> After you found one (say #3), remember all you PVs belonging to that VG and run
> 
> pvcreate -yff /dev/AllYourPVsBelongingToA
> for pv in /dev/AllYourPVsBelongingToA
> do
>    vgcfgrestore -f /etc/lvmconf/A.conf.3.old -n A $pv
> done
> vgscan
> vgchange -ay A

Unfortunately, that doesn't work because I  had a cron job that was creating
daily snapshots (in an attempt to emulate ONTAP's snapshot's feature), so my
10 backups contain snapshot LVs

If the backups were text, I could edit them and remove the snapshot info in
there, but with a binary backup, I don't know how to do that.

If I  know how big  my /var and  /usr partitions were  and in which  order I
created them (vgcfgrestore -ll shows this  anyway), can I recreate the VG on
top of itself without losing the data in the the LVs?
Is there another way to recover from this?

More importantly, is this a known bug? Can I give info to help fix it?

Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

Re: [linux-lvm] vgchange -a y /dev/System fails

[Heinz J . Mauelshagen]

On Tue, Jan 29, 2002 at 01:57:38PM -0800, Marc MERLIN wrote:
> [Adding a Cc to lvm-devel, since I apparently hit a bug in the code. Please
> remove or edit as appropriate]
> 
> On Mon, Jan 28, 2002 at 10:42:58AM +0100, Heinz J . Mauelshagen wrote:
> > > It fails and says:
> > > parameter error setting up snapshot copy on write exception
> > > 
> > > So yes, I have  snapshots in there, but I can't delete  the snapshots if the
> > > volume isn't active, and I can't activate the volume because of a snapshot.
> > > 
> > > To make things worse, /dev/System contains /usr and /var.
> > > 
> > > The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17 with
> > > 1.0.1rc4-ish
> > 
> > I am afraid you need to figure out the last metadata backup in /etc/lvmconf/
> > which *doesn't* contain the flaky snapshot(s).
> > 
> > You didn't mention the name of you VG so let's say it was 'A'.
> >
> > You can do that with "vgcfgrestore -f /etc/lvmconf/A.conf.1.old -n A -ll".
> > Replace the '1' in the metadata backup filename above with 2, 3 and so on
> > (you should have a default of up to 9).
> > After you found one (say #3), remember all you PVs belonging to that VG and run
> > 
> > pvcreate -yff /dev/AllYourPVsBelongingToA
> > for pv in /dev/AllYourPVsBelongingToA
> > do
> >    vgcfgrestore -f /etc/lvmconf/A.conf.3.old -n A $pv
> > done
> > vgscan
> > vgchange -ay A
> 
> Unfortunately, that doesn't work because I  had a cron job that was creating
> daily snapshots (in an attempt to emulate ONTAP's snapshot's feature), so my
> 10 backups contain snapshot LVs

Well, in this case you might still have a consistent backup with snapshots
which you can use like mentioned above.

> 
> If the backups were text, I could edit them and remove the snapshot info in
> there, but with a binary backup, I don't know how to do that.
> 
> If I  know how big  my /var and  /usr partitions were  and in which  order I
> created them (vgcfgrestore -ll shows this  anyway), can I recreate the VG on
> top of itself without losing the data in the the LVs?
> Is there another way to recover from this?
> 
> More importantly, is this a known bug? Can I give info to help fix it?

We get ocassional problem reports like yours.
Any syslog or console output which is related to a snapshot flaw would help.

> 
> Marc
> -- 
> Microsoft is to operating systems & security ....
>                                       .... what McDonalds is to gourmet cooking
>   
> Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 

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] vgchange -a y /dev/System fails

[Marc MERLIN]

On Wed, Jan 30, 2002 at 01:12:05PM +0100, Heinz J . Mauelshagen wrote:
> > Unfortunately, that doesn't work because I  had a cron job that was creating
> > daily snapshots (in an attempt to emulate ONTAP's snapshot's feature), so my
> > 10 backups contain snapshot LVs
> 
> Well, in this case you might still have a consistent backup with snapshots
> which you can use like mentioned above.

I tried a few and none seemed to work, but I restored one from before I made
snapshots and recovered from this.
I now have my system back (/usr, /var)

As  for  /home,  as soon  as  I  tried  to  access it  (e2fsck,  tune2fs  -O
^has_journal, or whatever), the process hung in "wait_o" according to ps.

I tried to  restore the VG to  its state before I added  snapshots, but that
only made it go away :-(
 
mechastreisand:/etc/lvmconf# vgchange -a n /dev/BigAssLVMRaid 
vgchange -- volume group "BigAssLVMRaid" successfully deactivated

mechastreisand:/etc/lvmconf# vgcfgrestore -f /etc/lvmconf/GoodBigAssLVMRaid.conf -n BigAssLVMRaid -ll 
--- Volume group ---
VG Name               BigAssLVMRaid
VG Access             read/write
VG Status             NOT available/resizable
VG #                  1
MAX LV                255
Cur LV                1
Open LV               0
MAX LV Size           1.00 TB
Max PV                255
Cur PV                9
Act PV                9
VG Size               868.69 GB
PE Size               64.00 MB
Total PE              13899
Alloc PE / Size       13120 / 820.00 GB
Free  PE / Size       779 / 48.69 GB
VG UUID               VgCZzh-ldUN-Ix1n-mZL3-EWsj-ATke-7Y4RQj

--- Logical volume ---
LV Name                /dev/BigAssLVMRaid/BigAssLVMPartition
VG Name                BigAssLVMRaid
LV Write Access        read/write
LV Status              NOT available
LV #                   1
# open                 0
LV Size                820.00 GB
Current LE             13120
Allocated LE           13120
Allocation             next free
Read ahead sectors     32
Block device           58:0


--- Physical volume ---
PV Name               /dev/rd/disc0/part7
VG Name               BigAssLVMRaid
PV Size               72.31 GB / NOT usable 121.73 MB [LVM: 128.00 KB]
PV#                   1
PV Status             NOT available
Allocatable           yes (but full)
Cur LV                1
PE Size (KByte)       65536
Total PE              1155
Free PE               0
Allocated PE          1155
PV UUID               Eriv40-q8V1-jgaS-ZXKh-8xwm-Wq7V-H9tVrl
(...)
mechastreisand:/etc/lvmconf# vgcfgrestore -f /etc/lvmconf/GoodBigAssLVMRaid.conf -n BigAssLVMRaid -ll | grep PV | grep rd | sed "s/.*\/dev/\/dev/"
/dev/rd/disc0/part7
/dev/rd/disc1/part1
/dev/rd/disc2/part1
/dev/rd/c1d0p1
/dev/rd/c1d1p1
/dev/rd/c2d0p1
/dev/rd/c2d1p1
/dev/rd/c3d0p1
/dev/rd/c3d1p1
mechastreisand:/etc/lvmconf# vgcfgrestore -f /etc/lvmconf/GoodBigAssLVMRaid.conf -n BigAssLVMRaid -ll | grep PV | grep rd | sed "s/.*\/dev/\/dev/" > /tmp/partlist
mechastreisand:/etc/lvmconf# pvcreate -yff `cat /tmp/partlist `
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/disc0/part7" of volume group "BigAssLVMRaid"
pvcreate -- removing lvmtab entry
pvcreate -- physical volume "/dev/rd/disc0/part7" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/disc1/part1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/disc1/part1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/disc2/part1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/disc2/part1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/c1d0p1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/c1d0p1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/c1d1p1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/c1d1p1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/c2d0p1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/c2d0p1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/c2d1p1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/c2d1p1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/c3d0p1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/c3d0p1" successfully created
pvcreate -- WARNING: forcing physical volume creation on "/dev/rd/c3d1p1" of volume group "BigAssLVMRaid"
pvcreate -- physical volume "/dev/rd/c3d1p1" successfully created

mechastreisand:/etc/lvmconf# for pv in `cat /tmp/partlist `
> do
> vgcfgrestore -f /etc/lvmconf/GoodBigAssLVMRaid.conf -n BigAssLVMRaid $pv
> done
vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/disc0/part7"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/disc1/part1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/disc2/part1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/c1d0p1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/c1d1p1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/c2d0p1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/c2d1p1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/c3d0p1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical volume "/dev/rd/c3d1p1"
vgcfgrestore -- you may not have an actual backup of restored volume group "BigAssLVMRaid"

mechastreisand:/etc/lvmconf# vgscan 
vgscan -- reading all physical volumes (this may take a while...)
vgscan -- found active volume group "System"
vgscan -- "/etc/lvmtab" and "/etc/lvmtab.d" successfully created
vgscan -- WARNING: This program does not do a VGDA backup of your volume group


-> My BigAssLVMRaid VG is gone :-(

Did I just lose my data?

Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

Re: [linux-lvm] vgchange -a y /dev/System fails

[Heinz J . Mauelshagen]

On Wed, Jan 30, 2002 at 07:04:17PM -0800, Marc MERLIN wrote:
> On Wed, Jan 30, 2002 at 01:12:05PM +0100, Heinz J . Mauelshagen wrote:
> > > Unfortunately, that doesn't work because I  had a cron job that was creating
> > > daily snapshots (in an attempt to emulate ONTAP's snapshot's feature), so my
> > > 10 backups contain snapshot LVs
> > 
> > Well, in this case you might still have a consistent backup with snapshots
> > which you can use like mentioned above.
> 
> I tried a few and none seemed to work, but I restored one from before I made
> snapshots and recovered from this.
> I now have my system back (/usr, /var)
> 
> As  for  /home,  as soon  as  I  tried  to  access it  (e2fsck,  tune2fs  -O
> ^has_journal, or whatever), the process hung in "wait_o" according to ps.
> 
> I tried to  restore the VG to  its state before I added  snapshots, but that
> only made it go away :-(
>  
> mechastreisand:/etc/lvmconf# vgchange -a n /dev/BigAssLVMRaid 

<SNIP>

> -> My BigAssLVMRaid VG is gone :-(
> 
> Did I just lose my data?

If you really tried all your metadata backups unsuccessfully, I am afraid so :-(
Some users reported snapshot problems with the recent 2.4.x kernels.
They are not 100% reliable due to VM issues being sorted out by the
kernel community hopefully which will hopefully be integrated in 2.4.18.

> 
> Marc
> -- 
> Microsoft is to operating systems & security ....
>                                       .... what McDonalds is to gourmet cooking
>   
> Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 

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] vgchange -a y /dev/System fails

[Marc MERLIN]

On Thu, Jan 31, 2002 at 12:14:17PM +0100, Heinz J . Mauelshagen wrote:
> > -> My BigAssLVMRaid VG is gone :-(
> > 
> > Did I just lose my data?
> 
> If you really tried all your  metadata backups unsuccessfully, I am afraid
> so  :-(  

I tried other ones, but every time, I'm getting:
vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical
volume "/dev/rd/disc0/part7"
vgcfgrestore -- you may not have an actual backup of restored volume group
"BigAssLVMRaid"

Am I supposed to give a special flag to vgcfgrestore, or maybe pvcreate?
(I did not use the default physicalextentsize)

As a reminder, I am currently running this command:
FILE=/etc/lvmconf/BigAssLVMRaid.conf.2.old; vgcfgrestore -f $FILE -n BigAssLVMRaid -ll; read a; pvcreate -yff `cat /tmp/partlist`; for pv in `cat /tmp/partlist `; do vgcfgrestore -f $FILE -n BigAssLVMRaid $pv; done; vgscan

Short of all this, can I re-create the VG the same way I did the first
time, and hope to get my filesytem again if I recreate the same LVs?
(like I can do with software raid)

This VG was working fine before I restored the other one (active and all)

> Some users reported snapshot problems with the recent 2.4.x kernels.  They
> are not  100% reliable  due to VM  issues being sorted  out by  the kernel
> community hopefully which will hopefully be integrated in 2.4.18.

Ok. I guess I'll know to steer clear  from snapshots in the meantime, but if
LVM doesn't even allow  me to recover a VG that  hasn't been overwriten with
random data, I don't know how much data I'm willing to trust to it...

Either way, thanks for your help.

Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

Re: [linux-lvm] vgchange -a y /dev/System fails

[Marc MERLIN]

On Fri, Feb 01, 2002 at 01:09:28PM -0800, Marc MERLIN wrote:

I wanted to thank Heinz publically for going beyond the call of duty, and
helping until the problem was fixed.
I thought I'd give a few lines to explain the resolution for the benefit of
other list members and the archives.

> On Thu, Jan 31, 2002 at 12:14:17PM +0100, Heinz J . Mauelshagen wrote:
> > > -> My BigAssLVMRaid VG is gone :-(
> > > 
> > > Did I just lose my data?
> > 
> > If you really tried all your  metadata backups unsuccessfully, I am afraid
> > so  :-(  
> 
> I tried other ones, but every time, I'm getting:
> vgcfgrestore -- VGDA for "BigAssLVMRaid" successfully restored to physical
> volume "/dev/rd/disc0/part7"
> vgcfgrestore -- you may not have an actual backup of restored volume group
> "BigAssLVMRaid"

The reason why vgchange -a -y didn't pick up  my VG is due to a small bug in
vgcfgrestore, which  only restored the  UUIDs of 8 out  of my 9  PVs (quoted
from memory, sorry if I got this slightly wrong).
Either way, the fix it to upgrade to  a newer version of the lvm tools (from
CVS or the next snapshot)

> This VG was working fine before I restored the other one (active and all)

The reason why I tried to restore it in the first place is that any kind of
partition access on that VG would hang the machine.
The problem turned out to be the crappy VM in 2.4.17, and not LVM.

Actually 2.4.14 also had a few  issues fscking the 800G partition, and froze
when  fsck was  going to  write  the changes  to  the partition  (and its  4
snapshots)
It probably would have completed had I deleted the snapshots, but I ended up
going  back to  2.4.7, and  after an  8H fsck,  the partition  was mountable
again.

Again, thanks to Heinz for his invaluable help.

Marc
-- 
Microsoft is to operating systems & security ....
                                      .... what McDonalds is to gourmet cooking
  
Home page: http://marc.merlins.org/   |   Finger marc_f@merlins.org for PGP key

Re: [linux-lvm] vgchange -a y /dev/System fails

[Adrian Head]

A little while ago I had the same problems - the only way i could get around 
it  was using 2.4.16-xfs.  I was using 2.4.17-xfs+lvm (1.0.1) and when a 
snapshot failed I couldn't remove it and when I rebooted the system would 
hang.  By having a 2.4.16-xfs kernel with lvm-1.0.1rc4(ish) it would fix the 
problems on the fly during bootup and all was happy.  YMMV  At the time I did 
ask the mailling list why 2.4.16-xfs was more robust than 2.4.17-xfs+lvm; 
however, I never recieved an answer.

Heady

On Sat, 26 Jan 2002 09:35, you wrote:
> It fails and says:
> parameter error setting up snapshot copy on write exception
>
> So yes, I have  snapshots in there, but I can't delete  the snapshots if
> the volume isn't active, and I can't activate the volume because of a
> snapshot.
>
> To make things worse, /dev/System contains /usr and /var.
>
> The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17
> with 1.0.1rc4-ish
>
> What's the magic override I need to use to make the volume active again?
>
> Thanks
> Marc

-- 
Adrian Head

(Public Key available on request.)

Re: [linux-lvm] vgchange -a y /dev/System fails

[Heinz J . Mauelshagen]

On Wed, Jan 30, 2002 at 10:24:35AM +1000, Adrian Head wrote:
> A little while ago I had the same problems - the only way i could get around 
> it  was using 2.4.16-xfs.  I was using 2.4.17-xfs+lvm (1.0.1) and when a 
> snapshot failed I couldn't remove it and when I rebooted the system would 
> hang.  By having a 2.4.16-xfs kernel with lvm-1.0.1rc4(ish) it would fix the 
> problems on the fly during bootup and all was happy.  YMMV  At the time I did 
> ask the mailling list why 2.4.16-xfs was more robust than 2.4.17-xfs+lvm; 
> however, I never recieved an answer.

Well, as linux-kernel reports there's still problems with 2.4 up to .17 in
the vm subsystem which show up on heavier load or on long running systems.
Andrea Arcangeli has some fixes but I don't know if they already have been
ported to 2.4.18-pre jet.

> 
> Heady
> 
> On Sat, 26 Jan 2002 09:35, you wrote:
> > It fails and says:
> > parameter error setting up snapshot copy on write exception
> >
> > So yes, I have  snapshots in there, but I can't delete  the snapshots if
> > the volume isn't active, and I can't activate the volume because of a
> > snapshot.
> >
> > To make things worse, /dev/System contains /usr and /var.
> >
> > The snapshots were created with 1.0.1rc4 on 2.4.14 and I now run 2.4.17
> > with 1.0.1rc4-ish
> >
> > What's the magic override I need to use to make the volume active again?
> >
> > Thanks
> > Marc
> 
> -- 
> Adrian Head
> 
> (Public Key available on request.)
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 

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