re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[Greg Freemyer]

I agree, their seems to be a bug in LVM.

I am running XFS Version 1.2pre3 based on the 2.4.19 kernel and I cannot extend a LV with XFS fs on top.

In particular this works:

>>>
xfs_freeze -f /data                                                  (I don't have the patch installed)
lvcreate --snapshot -L 100M --name data_snap /dev/VG/data 
xfs_freeze -u /data                                                 (I don't have the patch installed)
mount -t xfs -o ro,nouuid /dev/VG/data_snap /data_snap
<<<

And this does not.

>>>
xfs_freeze -f /data                                                  (I don't have the patch installed)
lvcreate --snapshot -L 100M --name data_snap /dev/VG/data 
xfs_freeze -u /data                                                 (I don't have the patch installed)

lvdisplay /dev/VG/data_snap                                   (verified 0% used)
dd if=/dev/zero of=/data/dummy bs=1M count=75     (create a 75 Meg dummy file)
lvdisplay /dev/VG/data_snap                                   (verified approx 75% used)
lvextend -L +100M /dev/VG/data_snap          
lvdisplay /dev/VG/data_snap                                   (verified approx 37% used)

#  Everything looks fine, so I try to mount it

mount -t xfs -o ro,nouuid /dev/VG/data_snap /data_snap
mount: wrong fs type, bad option, bad superblock on /dev/TruStore-Data/data_snap,
       or too many mounted file systems
<<<


Greg Freemyer
The Norcross Group

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[Sean Oh]

What I found out testing with a few FS available for Linux is that only the
snapshot for ext2 is extendable. I and extend ext2 snapshots with 'lvextend'
command and it keep staying mountable.

Snapshots for LV on Reiserfs, XFS and even ext3 are not
extendable(mountable).

Very strange, since LVM snapshot does not hold FS. Am I right?


----- Original Message -----
From: "Greg Freemyer" <freemyer@NorcrossGroup.com>
To: "LVM Mailing list" <linux-lvm@sistina.com>
Cc: "xfs mailing list" <linux-xfs@oss.sgi.com>
Sent: Tuesday, January 14, 2003 6:47 AM
Subject: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?



I agree, their seems to be a bug in LVM.

I am running XFS Version 1.2pre3 based on the 2.4.19 kernel and I cannot
extend a LV with XFS fs on top.

In particular this works:

>>>
xfs_freeze -f /data                                                  (I
don't have the patch installed)
lvcreate --snapshot -L 100M --name data_snap /dev/VG/data
xfs_freeze -u /data                                                 (I don't
have the patch installed)
mount -t xfs -o ro,nouuid /dev/VG/data_snap /data_snap
<<<

And this does not.

>>>
xfs_freeze -f /data                                                  (I
don't have the patch installed)
lvcreate --snapshot -L 100M --name data_snap /dev/VG/data
xfs_freeze -u /data                                                 (I don't
have the patch installed)

lvdisplay /dev/VG/data_snap                                   (verified 0%
used)
dd if=/dev/zero of=/data/dummy bs=1M count=75     (create a 75 Meg dummy
file)
lvdisplay /dev/VG/data_snap                                   (verified
approx 75% used)
lvextend -L +100M /dev/VG/data_snap
lvdisplay /dev/VG/data_snap                                   (verified
approx 37% used)

#  Everything looks fine, so I try to mount it

mount -t xfs -o ro,nouuid /dev/VG/data_snap /data_snap
mount: wrong fs type, bad option, bad superblock on
/dev/TruStore-Data/data_snap,
       or too many mounted file systems
<<<


Greg Freemyer
The Norcross Group


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

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[Dale J. Stephenson]

On Thursday, January 16, 2003, at 06:38  PM, Sean Oh wrote:

> What I found out testing with a few FS available for Linux is that 
> only the
> snapshot for ext2 is extendable. I and extend ext2 snapshots with 
> 'lvextend'
> command and it keep staying mountable.
>
> Snapshots for LV on Reiserfs, XFS and even ext3 are not
> extendable(mountable).
>
> Very strange, since LVM snapshot does not hold FS. Am I right?
>
At the FS layer, it should appear to be the same size as before (the 
size of the original).  I've had no trouble at all reducing or 
extending XFS snapshot volumes while they were mounted, but I don't 
think I've tried it unmounted.

Dale J. Stephenson
dalestephenson@mac.com

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[Sean Oh]

----- Original Message -----
From: "Dale J. Stephenson" <dalestephenson@mac.com>
To: <linux-lvm@sistina.com>
Sent: Friday, January 17, 2003 8:55 AM
Subject: Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?
> At the FS layer, it should appear to be the same size as before (the
> size of the original).  I've had no trouble at all reducing or
> extending XFS snapshot volumes while they were mounted, but I don't
> think I've tried it unmounted.

Well, my system is rebooting once in awhile, so the extended snapshot shoud
be able to be remounted.

Does anyone knows some solution for this?

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[jon+lvm]

On Fri, Jan 17, 2003 at 08:38:51AM +0900, Sean Oh wrote:
> What I found out testing with a few FS available for Linux is that only the
> snapshot for ext2 is extendable. I and extend ext2 snapshots with 'lvextend'
> command and it keep staying mountable.
> 
> Snapshots for LV on Reiserfs, XFS and even ext3 are not
> extendable(mountable).
> 
> Very strange, since LVM snapshot does not hold FS. Am I right?

i have successfully extented a ext3 snapshot. I did have to specify the
mount as READONLY. You did do that too, didnt you ?



JonB

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[Sean Oh]

----- Original Message -----
From: <jon+lvm@silicide.dk>
To: <linux-lvm@sistina.com>
Sent: Friday, January 17, 2003 7:02 PM
Subject: Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?
> >
> > Very strange, since LVM snapshot does not hold FS. Am I right?
>
> i have successfully extented a ext3 snapshot. I did have to specify the
> mount as READONLY. You did do that too, didnt you ?
>

Yes, I specified the ro option in ext3 snapshots. But it is still not
mountable, giving the error 'Cant write to read-only device' and 'EXT3-FS:
write access unavailable, cannot proceed'

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[jon+lvm]

On Fri, Jan 17, 2003 at 08:31:34PM +0900, Sean Oh wrote:
> 
> ----- Original Message -----
> From: <jon+lvm@silicide.dk>
> To: <linux-lvm@sistina.com>
> Sent: Friday, January 17, 2003 7:02 PM
> Subject: Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?
> > >
> > > Very strange, since LVM snapshot does not hold FS. Am I right?
> >
> > i have successfully extented a ext3 snapshot. I did have to specify the
> > mount as READONLY. You did do that too, didnt you ?
> >
> 
> Yes, I specified the ro option in ext3 snapshots. But it is still not
> mountable, giving the error 'Cant write to read-only device' and 'EXT3-FS:
> write access unavailable, cannot proceed'

strange.



JonB

Re: re[2]: [linux-lvm] [Q] LVM snapshot volume extendable?

[Jose Luis Domingo Lopez]

On Friday, 17 January 2003, at 20:31:34 +0900,
Sean Oh wrote:

> Yes, I specified the ro option in ext3 snapshots. But it is still not
> mountable, giving the error 'Cant write to read-only device' and 'EXT3-FS:
> write access unavailable, cannot proceed'
> 
Did you try to mount the ext3 snapshot not only in read-only mode, but
_also_ without replaying the journal (option -o noload). I can't verify
this, because my kernel is compiled without the required "vfs-lock patch".

-- 
Jose Luis Domingo Lopez
Linux Registered User #189436     Debian Linux Sid (Linux 2.4.20-xfs)

[linux-lvm] Kernel PANIC after partition change on RedHat 8.0

[Yuliy Minchev]

Hi!

I have installed Red Hat Linux 8.0. I've made volume group (rootvg)
from /dev/md1 (which is raid1). I've put my /boot on /dev/md0.

As you already have noticed - I have two disks.
After installation I'm making new partiotion /dev/sdb5 i.e.
And if I change its to type to 8e after reboot kernel panics.
If I change type of this partition to something different - the system 
boots fine.

I cannot understand where can be the problem.
I have to say that I made the same configuration on this machine and 
everything was ok. But now I cannot see any difference.

Any ideas?

TIA

yuliy

-- 
  Yuliy Minchev, 
  UNIX Administrator

Re: [linux-lvm] Kernel PANIC after partition change on RedHat 8.0

[Steven Lembark]

> I have installed Red Hat Linux 8.0. I've made volume group (rootvg)
> from /dev/md1 (which is raid1). I've put my /boot on /dev/md0.
>
> As you already have noticed - I have two disks.
> After installation I'm making new partiotion /dev/sdb5 i.e.
> And if I change its to type to 8e after reboot kernel panics.
> If I change type of this partition to something different - the system
> boots fine.
>
> I cannot understand where can be the problem.
> I have to say that I made the same configuration on this machine and
> everything was ok. But now I cannot see any difference.

sdb5 => sub-partition. Are you accidentally carving out
a piece of an existing PV? For example, if you had partitioned
the disk into /dev/sdb[1234] when it was built with sdb4 taking
up most of the device as 8e (i.e., for LVM), then adding sdb5
w/in the area of sdb4 would certianly break things -- and scramble
your LV's during pvcreate on the new partition.


--
Steven Lembark                               2930 W. Palmer
Workhorse Computing                       Chicago, IL 60647
                                            +1 773 252 1080

Re: [linux-lvm] Kernel PANIC after partition change on RedHat 8.0

[Yuliy Minchev]

On Sat, 18 Jan 2003, Steven Lembark wrote:

> 
> > I have installed Red Hat Linux 8.0. I've made volume group (rootvg)
> > from /dev/md1 (which is raid1). I've put my /boot on /dev/md0.
> >
> > As you already have noticed - I have two disks.
> > After installation I'm making new partiotion /dev/sdb5 i.e.
> > And if I change its to type to 8e after reboot kernel panics.
> > If I change type of this partition to something different - the system
> > boots fine.
> >
> > I cannot understand where can be the problem.
> > I have to say that I made the same configuration on this machine and
> > everything was ok. But now I cannot see any difference.
> 
> sdb5 => sub-partition. Are you accidentally carving out
> a piece of an existing PV? For example, if you had partitioned
> the disk into /dev/sdb[1234] when it was built with sdb4 taking
> up most of the device as 8e (i.e., for LVM), then adding sdb5
> w/in the area of sdb4 would certianly break things -- and scramble
> your LV's during pvcreate on the new partition.

I have

/dev/sdb1 /boot (member of md0,raid1)
/dev/sdb1 member of md1, raid1, on md1 is my volume group
/dev/sdb3 fat32
/dev/sbd4 extend partion
/dev/sdb5 is defined as 8e within sdb4

Sorry, but I did not understand your point :(

yuliy

> 
> 
> --
> Steven Lembark                               2930 W. Palmer
> Workhorse Computing                       Chicago, IL 60647
>                                             +1 773 252 1080
> 
> _______________________________________________
> 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/
> 

-- 
  Yuliy Minchev, 
  UNIX Administrator

Re: [linux-lvm] Kernel PANIC after partition change on RedHat 8.0

[Steven Lembark]

-- Yuliy Minchev <yuliy@mobiltel.bg>

> On Sat, 18 Jan 2003, Steven Lembark wrote:
>
>>
>> > I have installed Red Hat Linux 8.0. I've made volume group (rootvg)
>> > from /dev/md1 (which is raid1). I've put my /boot on /dev/md0.
>> >
>> > As you already have noticed - I have two disks.
>> > After installation I'm making new partiotion /dev/sdb5 i.e.
>> > And if I change its to type to 8e after reboot kernel panics.
>> > If I change type of this partition to something different - the system
>> > boots fine.
>> >
>> > I cannot understand where can be the problem.
>> > I have to say that I made the same configuration on this machine and
>> > everything was ok. But now I cannot see any difference.
>>
>> sdb5 => sub-partition. Are you accidentally carving out
>> a piece of an existing PV? For example, if you had partitioned
>> the disk into /dev/sdb[1234] when it was built with sdb4 taking
>> up most of the device as 8e (i.e., for LVM), then adding sdb5
>> w/in the area of sdb4 would certianly break things -- and scramble
>> your LV's during pvcreate on the new partition.
>
> I have
>
> /dev/sdb1 /boot (member of md0,raid1)
> /dev/sdb1 member of md1, raid1, on md1 is my volume group
> /dev/sdb3 fat32
> /dev/sbd4 extend partion
> /dev/sdb5 is defined as 8e within sdb4
>
> Sorry, but I did not understand your point :(

One thing people have zapped themselves with in the past
is something like:

/dev/sdb4 == one partition
/dev/sdb5 == another partition whose space is
             contained w/in /dev/sdb4

	pvcreate -ff /dev/sdb4;
	vgcreate vg00 /dev/sdb4;
	lvcretate ... vg00;

	<later in time>

	pvcreate -ff /dev/sdb5;
	vgcreate vg01 /dev/sdb5;
	lvcreate ... vg01;

at this point the creation of vg01 has overwritten
data in vg00 becuase the partition /dev/sda5 uses
disk realestate contained within /dev/sda4. This is
a matter purely caused by having moultiple partitions
w/in the same spindle.

That is why I wondered if you had made the mistake
of creating a partition w/in an already-existing
partition which was used as a PV at the time when
your new partition was created and used.



--
Steven Lembark                               2930 W. Palmer
Workhorse Computing                       Chicago, IL 60647
                                            +1 773 252 1080

Re: [linux-lvm] Kernel PANIC after partition change on RedHat 8.0

[Yuliy Minchev]

On Sun, 19 Jan 2003, Steven Lembark wrote:

> 
> 
> -- Yuliy Minchev <yuliy@mobiltel.bg>
> 
> > On Sat, 18 Jan 2003, Steven Lembark wrote:
> >
> >>
> >> > I have installed Red Hat Linux 8.0. I've made volume group (rootvg)
> >> > from /dev/md1 (which is raid1). I've put my /boot on /dev/md0.
> >> >
> >> > As you already have noticed - I have two disks.
> >> > After installation I'm making new partiotion /dev/sdb5 i.e.
> >> > And if I change its to type to 8e after reboot kernel panics.
> >> > If I change type of this partition to something different - the system
> >> > boots fine.
> >> >
> >> > I cannot understand where can be the problem.
> >> > I have to say that I made the same configuration on this machine and
> >> > everything was ok. But now I cannot see any difference.
> >>
> >> sdb5 => sub-partition. Are you accidentally carving out
> >> a piece of an existing PV? For example, if you had partitioned
> >> the disk into /dev/sdb[1234] when it was built with sdb4 taking
> >> up most of the device as 8e (i.e., for LVM), then adding sdb5
> >> w/in the area of sdb4 would certianly break things -- and scramble
> >> your LV's during pvcreate on the new partition.
> >
> > I have
> >
> > /dev/sdb1 /boot (member of md0,raid1)
> > /dev/sdb1 member of md1, raid1, on md1 is my volume group
> > /dev/sdb3 fat32
> > /dev/sbd4 extend partion
> > /dev/sdb5 is defined as 8e within sdb4
> >
> > Sorry, but I did not understand your point :(
> 
> One thing people have zapped themselves with in the past
> is something like:
> 
> /dev/sdb4 == one partition
> /dev/sdb5 == another partition whose space is
>              contained w/in /dev/sdb4
> 
> 	pvcreate -ff /dev/sdb4;
> 	vgcreate vg00 /dev/sdb4;
> 	lvcretate ... vg00;
> 
> 	<later in time>
> 
> 	pvcreate -ff /dev/sdb5;
> 	vgcreate vg01 /dev/sdb5;
> 	lvcreate ... vg01;
> 
> at this point the creation of vg01 has overwritten
> data in vg00 becuase the partition /dev/sda5 uses
> disk realestate contained within /dev/sda4. This is
> a matter purely caused by having moultiple partitions
> w/in the same spindle.
> 
> That is why I wondered if you had made the mistake
> of creating a partition w/in an already-existing
> partition which was used as a PV at the time when
> your new partition was created and used.

No, no...
The case is not like this. There is no such overlapping here.

I created my volume group during RH8 installation.
After that I manually created /dev/sdb4 and within /dev/sbd5 with fdisk.
And when I change the type of /dev/sbd5 to 8e in order to be able
to run pvcreate ot it and to add it to my volume groups - the
systems hangs during kernel loading. In fact, the system stops to 
recognise my volume group. That's why it hangs.

yuliy


> 
> 
> 
> --
> Steven Lembark                               2930 W. Palmer
> Workhorse Computing                       Chicago, IL 60647
>                                             +1 773 252 1080
> 
> _______________________________________________
> 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/
> 

-- 
  Yuliy Minchev, 
  UNIX Administrator

Re: [linux-lvm] Kernel PANIC after partition change on RedHat 8.0

[Yuliy Minchev]

On Sun, 19 Jan 2003, Yuliy Minchev wrote:

I have to update the case.

I've tried to move /dev/sdb3 to the end of the hard drive.
After this, I recreated sdb45 on the same size free space, but starting at 
lower cyl (again after 1024).
And it works. Systems does not hang. I can run pvcreate, vgextend and so 
on...

After this I deleted again disk partition sdb45
and move /dev/sdb3 on its inital place.
I recreated again sdb45 and again system hangs.

This whole lot looks to strange to me :(
It should be like this. As long as I know - position on hard disk is not 
important at all.


> On Sun, 19 Jan 2003, Steven Lembark wrote:
> 
> > 
> > 
> > -- Yuliy Minchev <yuliy@mobiltel.bg>
> > 
> > > On Sat, 18 Jan 2003, Steven Lembark wrote:
> > >
> > >>
> > >> > I have installed Red Hat Linux 8.0. I've made volume group (rootvg)
> > >> > from /dev/md1 (which is raid1). I've put my /boot on /dev/md0.
> > >> >
> > >> > As you already have noticed - I have two disks.
> > >> > After installation I'm making new partiotion /dev/sdb5 i.e.
> > >> > And if I change its to type to 8e after reboot kernel panics.
> > >> > If I change type of this partition to something different - the system
> > >> > boots fine.
> > >> >
> > >> > I cannot understand where can be the problem.
> > >> > I have to say that I made the same configuration on this machine and
> > >> > everything was ok. But now I cannot see any difference.
> > >>
> > >> sdb5 => sub-partition. Are you accidentally carving out
> > >> a piece of an existing PV? For example, if you had partitioned
> > >> the disk into /dev/sdb[1234] when it was built with sdb4 taking
> > >> up most of the device as 8e (i.e., for LVM), then adding sdb5
> > >> w/in the area of sdb4 would certianly break things -- and scramble
> > >> your LV's during pvcreate on the new partition.
> > >
> > > I have
> > >
> > > /dev/sdb1 /boot (member of md0,raid1)
> > > /dev/sdb1 member of md1, raid1, on md1 is my volume group
> > > /dev/sdb3 fat32
> > > /dev/sbd4 extend partion
> > > /dev/sdb5 is defined as 8e within sdb4
> > >
> > > Sorry, but I did not understand your point :(
> > 
> > One thing people have zapped themselves with in the past
> > is something like:
> > 
> > /dev/sdb4 == one partition
> > /dev/sdb5 == another partition whose space is
> >              contained w/in /dev/sdb4
> > 
> > 	pvcreate -ff /dev/sdb4;
> > 	vgcreate vg00 /dev/sdb4;
> > 	lvcretate ... vg00;
> > 
> > 	<later in time>
> > 
> > 	pvcreate -ff /dev/sdb5;
> > 	vgcreate vg01 /dev/sdb5;
> > 	lvcreate ... vg01;
> > 
> > at this point the creation of vg01 has overwritten
> > data in vg00 becuase the partition /dev/sda5 uses
> > disk realestate contained within /dev/sda4. This is
> > a matter purely caused by having moultiple partitions
> > w/in the same spindle.
> > 
> > That is why I wondered if you had made the mistake
> > of creating a partition w/in an already-existing
> > partition which was used as a PV at the time when
> > your new partition was created and used.
> 
> No, no...
> The case is not like this. There is no such overlapping here.
> 
> I created my volume group during RH8 installation.
> After that I manually created /dev/sdb4 and within /dev/sbd5 with fdisk.
> And when I change the type of /dev/sbd5 to 8e in order to be able
> to run pvcreate ot it and to add it to my volume groups - the
> systems hangs during kernel loading. In fact, the system stops to 
> recognise my volume group. That's why it hangs.
> 
> yuliy
> 
> 
> > 
> > 
> > 
> > --
> > Steven Lembark                               2930 W. Palmer
> > Workhorse Computing                       Chicago, IL 60647
> >                                             +1 773 252 1080
> > 
> > _______________________________________________
> > 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/
> > 
> 
> 

-- 
  Yuliy Minchev, 
  UNIX Administrator