[linux-lvm] lvm 1 unable to boot

[linux-lvm] lvm 1 unable to boot

[Geoff Dolman]

Hi

I have a machine with the following partition structure:

/dev/sda1 /boot (ext3)
/dev/sda2 swap
/dev/sda3 LVM

The LVM (1.03/rh9) contains one PV and this has Volume00 in it which
contains lvs for slash, usr, var, /usr/local and so on...

I rebooted the machine for the first time in ages (same kernel
configuration as the last reboot and no changes - or very few).

The machine won't reboot - it panics because of a message (something)
like:

vgscan found inactive "Volume00"
Error 28 Unable to make /etc/lvmtab.d/Volume00/Volume00.tmp
vg_cfgbackup.c line 273

It then gives an error about pivot_root panic with the error unable to
mount /dev/Volume00/slash mount error 2 
(the same sort of error as if fstab was wrong but it isn't)


Here's what I've tried to fix this:

- boot into rescue mode: lvscan shows all lvs no problem
	checked pvscan, vgscan, lvscan - no problems
	fscked slash - okay
	checked configuration files, fstab, etc.
tried a reboot - another panic
- rescue mode again
	delete /etc/lvmtab and /etc/lvmtab.d
reboot - panic
- rescue: reinstall lvm rpm and kernel
panic
- make a new initrd.img
panic
...plus lots of other things that were probably irrelevant.

The machine *still* panics when booted normally but works fine in rescue
mode, with a chrooted slash.

Any ideas?

Thanks

Geoff Dolman
-- 
JDRF/WT Diabetes and Inflammation Laboratory
Cambridge Institute for Medical Research
University of Cambridge
http://www-gene.cimr.cam.ac.uk/todd/

Re: [linux-lvm] lvm 1 unable to boot

[Patrick Caulfield]

On Wed, Mar 31, 2004 at 03:59:39PM +0100, Geoff Dolman wrote:
> Hi
> 
> I have a machine with the following partition structure:
> 
> /dev/sda1 /boot (ext3)
> /dev/sda2 swap
> /dev/sda3 LVM
> 
> The LVM (1.03/rh9) contains one PV and this has Volume00 in it which
> contains lvs for slash, usr, var, /usr/local and so on...
> 
> I rebooted the machine for the first time in ages (same kernel
> configuration as the last reboot and no changes - or very few).
> 
> The machine won't reboot - it panics because of a message (something)
> like:
> 
> vgscan found inactive "Volume00"
> Error 28 Unable to make /etc/lvmtab.d/Volume00/Volume00.tmp
> vg_cfgbackup.c line 273
> 

Error 28 is ENOSPC - your initrd is too small to hold the metadata backups.
-- 

patrick

[linux-lvm] max LV size

[Alexander Lazarevich]

Hi,

Using lvm-1.0.3-15 on RHEL3-AS, kernel 2.4.21-9.0.1.ELsmp.

I've got two SCSI devices attached to the system, each one is 1.74TB in 
size. I want to make a single ext3 filesystem out of those devices. So I'm 
using LVM to create a logical volume (3.48TB) out of those devices:

Even though I set the PE size to 512MB ("vgcreate -s512 /dev/sda 
/dev/sdb"), and I can see the VG size is correct (3.18TB), the maximum LV 
size is still 2TB:

#vgdisplay
[root@xxxxxx root]# vgdisplay
--- Volume group ---
VG Name               test-vg
VG Access             read/write
VG Status             available/resizable
VG #                  0
MAX LV                256
Cur LV                0
Open LV               0
MAX LV Size           2 TB
Max PV                256
Cur PV                2
Act PV                2
VG Size               3.18 TB
PE Size               512 MB
Total PE              6510
Alloc PE / Size       0 / 0
Free  PE / Size       6510 / 3.18 TB
VG UUID               dU1Xjg-w0Y4-aeh2-yU4l-Wo7g-Lewx-r6l1cT
 
Later in the man pages, it says something briefly about a 2TB block device 
limit in linux 2.4:

#man vgcreate
"There is also (as of Linux 2.4) a kernel limitation of 2TB per block 
device."

Is an LV a "block device" and is this why I can't seem to create an LV 
bigger than 2TB? If this is the problem, then LVM2 wouldn't even help, 
would it? Is it really true that no one can create a non-softwareRAID 
filesystem bigger than 2TB on linux 2.4? That just can't be true. Can it?

Thanks in advance,

Alex
---                                                               ---
   Alex Lazarevich | Systems Administrator | Imaging Technology Group
    Beckman Institute | University of Illinois | www.itg.uiuc.edu
---                                                               ---

Re: [linux-lvm] lvm 1 unable to boot

[Martijn Schoemaker]

Hi,

You can set this as an 'append ramdisksize=<size>' in your
lilo/bootloader config, but this does not always seem to
work. Best way seems to increase this number in the kernel,
but this means you need to build a new kernel first, and for
that, well, you need to be able to boot :)

Don't know much about GRUB, but the append option seems the
best solution for you.

Cheers,
Martijn

Geoff Dolman wrote:
> On Wed, 2004-03-31 at 16:49, Patrick Caulfield wrote:
> 
>>On Wed, Mar 31, 2004 at 03:59:39PM +0100, Geoff Dolman wrote:
>>
>>>Hi
>>>
>>>I have a machine with the following partition structure:
>>>
>>>/dev/sda1 /boot (ext3)
>>>/dev/sda2 swap
>>>/dev/sda3 LVM
>>>
>>>The LVM (1.03/rh9) contains one PV and this has Volume00 in it which
>>>contains lvs for slash, usr, var, /usr/local and so on...
>>>
>>>I rebooted the machine for the first time in ages (same kernel
>>>configuration as the last reboot and no changes - or very few).
>>>
>>>The machine won't reboot - it panics because of a message (something)
>>>like:
>>>
>>>vgscan found inactive "Volume00"
>>>Error 28 Unable to make /etc/lvmtab.d/Volume00/Volume00.tmp
>>>vg_cfgbackup.c line 273
>>>
>>
>>Error 28 is ENOSPC - your initrd is too small to hold the metadata backups.
> 
> 
> Thanks - but how do I fix this?
> 
> cheers
> 
> Geoff
> 

-- 
There's someone in my head, but it's not me.
--- Pink Floyd

Re: [linux-lvm] lvm 1 unable to boot

[Geoff Dolman]

On Wed, 2004-03-31 at 16:49, Patrick Caulfield wrote:
> On Wed, Mar 31, 2004 at 03:59:39PM +0100, Geoff Dolman wrote:
> > Hi
> > 
> > I have a machine with the following partition structure:
> > 
> > /dev/sda1 /boot (ext3)
> > /dev/sda2 swap
> > /dev/sda3 LVM
> > 
> > The LVM (1.03/rh9) contains one PV and this has Volume00 in it which
> > contains lvs for slash, usr, var, /usr/local and so on...
> > 
> > I rebooted the machine for the first time in ages (same kernel
> > configuration as the last reboot and no changes - or very few).
> > 
> > The machine won't reboot - it panics because of a message (something)
> > like:
> > 
> > vgscan found inactive "Volume00"
> > Error 28 Unable to make /etc/lvmtab.d/Volume00/Volume00.tmp
> > vg_cfgbackup.c line 273
> > 
> 
> Error 28 is ENOSPC - your initrd is too small to hold the metadata backups.

Thanks - but how do I fix this?

cheers

Geoff

-- 
JDRF/WT Diabetes and Inflammation Laboratory
Cambridge Institute for Medical Research
University of Cambridge
http://www-gene.cimr.cam.ac.uk/todd/

Re : [linux-lvm] lvm 1 unable to boot

[Jean-Luc Coulon (f5ibh)]

[-- Attachment #1: Type: text/plain, Size: 498 bytes --]

Le 31.03.2004 22:42, Geoff Dolman a écrit :
>On Wed, 2004-03-31 at 16:49, Patrick Caulfield wrote:
>> On Wed, Mar 31, 2004 at 03:59:39PM +0100, Geoff Dolman wrote:
>> > Hi

[.. destructive compression ..]

>>
>> Error 28 is ENOSPC - your initrd is too small to hold the metadata
>backups.
>
>Thanks - but how do I fix this?

If you use lvmcreate_initrd, it calculates automatically the size of  
the filesystem underlaying the initrd.


--
		- Jean-Luc
>
>cheers
>
>Geoff
>

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [linux-lvm] max LV size

[Heinz Mauelshagen]

On Wed, Mar 31, 2004 at 01:33:23PM -0600, Alexander Lazarevich wrote:
> Hi,
> 
> Using lvm-1.0.3-15 on RHEL3-AS, kernel 2.4.21-9.0.1.ELsmp.
> 
> I've got two SCSI devices attached to the system, each one is 1.74TB in 
> size. I want to make a single ext3 filesystem out of those devices. So I'm 
> using LVM to create a logical volume (3.48TB) out of those devices:
> 
> Even though I set the PE size to 512MB ("vgcreate -s512 /dev/sda 
> /dev/sdb"), and I can see the VG size is correct (3.18TB), the maximum LV 
> size is still 2TB:
> 
> #vgdisplay
> [root@xxxxxx root]# vgdisplay
> --- Volume group ---
> VG Name               test-vg
> VG Access             read/write
> VG Status             available/resizable
> VG #                  0
> MAX LV                256
> Cur LV                0
> Open LV               0
> MAX LV Size           2 TB
> Max PV                256
> Cur PV                2
> Act PV                2
> VG Size               3.18 TB
> PE Size               512 MB
> Total PE              6510
> Alloc PE / Size       0 / 0
> Free  PE / Size       6510 / 3.18 TB
> VG UUID               dU1Xjg-w0Y4-aeh2-yU4l-Wo7g-Lewx-r6l1cT
>  
> Later in the man pages, it says something briefly about a 2TB block device 
> limit in linux 2.4:
> 
> #man vgcreate
> "There is also (as of Linux 2.4) a kernel limitation of 2TB per block 
> device."
> 
> Is an LV a "block device" and is this why I can't seem to create an LV 
> bigger than 2TB? If this is the problem, then LVM2 wouldn't even help, 
> would it? Is it really true that no one can create a non-softwareRAID 
> filesystem bigger than 2TB on linux 2.4? That just can't be true. Can it?

It can. That limitation is gone in Linux 2.6.

> 
> Thanks in advance,
> 
> Alex
> ---                                                               ---
>    Alex Lazarevich | Systems Administrator | Imaging Technology Group
>     Beckman Institute | University of Illinois | www.itg.uiuc.edu
> ---                                                               ---
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@redhat.com
> https://www.redhat.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                                 Red Hat GmbH
Consulting Development Engineer                   Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@RedHat.com                            +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] lvm 1 unable to boot

[Heinz Mauelshagen]

On Wed, Mar 31, 2004 at 09:42:44PM +0100, Geoff Dolman wrote:
> On Wed, 2004-03-31 at 16:49, Patrick Caulfield wrote:
> > On Wed, Mar 31, 2004 at 03:59:39PM +0100, Geoff Dolman wrote:
> > > Hi
> > > 
> > > I have a machine with the following partition structure:
> > > 
> > > /dev/sda1 /boot (ext3)
> > > /dev/sda2 swap
> > > /dev/sda3 LVM
> > > 
> > > The LVM (1.03/rh9) contains one PV and this has Volume00 in it which
> > > contains lvs for slash, usr, var, /usr/local and so on...
> > > 
> > > I rebooted the machine for the first time in ages (same kernel
> > > configuration as the last reboot and no changes - or very few).
> > > 
> > > The machine won't reboot - it panics because of a message (something)
> > > like:
> > > 
> > > vgscan found inactive "Volume00"
> > > Error 28 Unable to make /etc/lvmtab.d/Volume00/Volume00.tmp
> > > vg_cfgbackup.c line 273
> > > 
> > 
> > Error 28 is ENOSPC - your initrd is too small to hold the metadata backups.
> 
> Thanks - but how do I fix this?

Append the kernel boot parameter "ramdisk_size=" and give a size in kilobytes
(eg, 'ramdisk_size=16384' makes 16MB ram disks.

> 
> cheers
> 
> Geoff
> 
> -- 
> JDRF/WT Diabetes and Inflammation Laboratory
> Cambridge Institute for Medical Research
> University of Cambridge
> http://www-gene.cimr.cam.ac.uk/todd/
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@redhat.com
> https://www.redhat.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                                 Red Hat GmbH
Consulting Development Engineer                   Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@RedHat.com                            +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re : [linux-lvm] lvm 1 unable to boot

[Jean-Luc Coulon (f5ibh)]

[-- Attachment #1: Type: text/plain, Size: 1961 bytes --]

Le 01.04.2004 15:48, Heinz Mauelshagen a écrit :
>On Wed, Mar 31, 2004 at 09:42:44PM +0100, Geoff Dolman wrote:
>> On Wed, 2004-03-31 at 16:49, Patrick Caulfield wrote:
>> > On Wed, Mar 31, 2004 at 03:59:39PM +0100, Geoff Dolman wrote:
>> > > Hi
>> > >
>> > > I have a machine with the following partition structure:
>> > >
>> > > /dev/sda1 /boot (ext3)
>> > > /dev/sda2 swap
>> > > /dev/sda3 LVM
>> > >
>> > > The LVM (1.03/rh9) contains one PV and this has Volume00 in it
>which
>> > > contains lvs for slash, usr, var, /usr/local and so on...
>> > >
>> > > I rebooted the machine for the first time in ages (same kernel
>> > > configuration as the last reboot and no changes - or very few).
>> > >
>> > > The machine won't reboot - it panics because of a message
>(something)
>> > > like:
>> > >
>> > > vgscan found inactive "Volume00"
>> > > Error 28 Unable to make /etc/lvmtab.d/Volume00/Volume00.tmp
>> > > vg_cfgbackup.c line 273
>> > >
>> >
>> > Error 28 is ENOSPC - your initrd is too small to hold the metadata
>backups.
>>
>> Thanks - but how do I fix this?
>
>Append the kernel boot parameter "ramdisk_size=" and give a size in
>kilobytes
>(eg, 'ramdisk_size=16384' makes 16MB ram disks.

I don't think so. It is not the ramdisk size which has a problem but  
the filesystem created to support the initrd files (/dev, /etc,  
modules, scripts, ...). The size of this filesystem is fixed while  
creating the initrd.

At boot time, lvm want to write some files in /etc (lvmtab and lvmtab. 
d). While in the boot process and in the ramdisk, the /etc is part of  
this filesystem. If there is not enough room on it, lvm crashes.

lvmcreate_initrd computes the needed space.

I tried to mount via the loopback an initrd created with  
lvmcreate_initrd, add the stuff needed to have xfs and I got short on  
the filesystem.... The ramdisk was 8Mo and the initrd about 6Mb...

--
Regards
			- Jean-Luc

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Re : [linux-lvm] lvm 1 unable to boot

[Luca Berra]

Jean-Luc Coulon (f5ibh) wrote:

> Le 01.04.2004 15:48, Heinz Mauelshagen a �crit :
>>
>> Append the kernel boot parameter "ramdisk_size=" and give a size in
>> kilobytes
>> (eg, 'ramdisk_size=16384' makes 16MB ram disks.
> 
> 
> I don't think so. It is not the ramdisk size which has a problem but  
> the filesystem created to support the initrd files (/dev, /etc,  
> modules, scripts, ...). The size of this filesystem is fixed while  
> creating the initrd.
edit /sbin/mkinitrd and increase the size there.

L.

-- 
Luca Berra -- bluca@comedia.it