All of lore.kernel.org
 help / color / mirror / Atom feed
* [linux-lvm] Sorry to ask here ...
@ 2017-02-23 19:25 Georges Giralt
  2017-02-23 20:23 ` Xen
  2017-03-02  5:44 ` Eric Ren
  0 siblings, 2 replies; 4+ messages in thread
From: Georges Giralt @ 2017-02-23 19:25 UTC (permalink / raw)
  To: LVM general discussion and development

Hello !

I'm sorry to ask for help here, but I'm lost in the middle of nowhere ...

The context :

A PC hardware with an UEFI "BIOS" set to legacy boot only.

3 disks with 2 partitions each. The 3 first partitions are set in 
software mirror (md0) and used for /boot (Ext4) .The second 3 partitions 
are used as a software mirror and the resulting md1 is the sole PV of a 
vg0 group onto which 8 logical volumes are set. This is used to run 
Ubuntu 16.04.2 software (after many upgrades...).   *

This setup has worked for years and seen changes in disk sizes, main 
board, etc...

Lastly, I bet the cause is an upgrade of some packages, the machine 
refuses to boot. It drops onto initramfs shell because the vg0 volume 
group is not activated. Launching lvm and doing a vgchange -a y does the 
trick and the system boots fine afterwards.

I've searched everything I can (but maybe the setup was so rock solid I 
became lazy or unconscious) and tried to make again the initramfs with 
no luck.

So every leads or advice would be greatly appreciated. Please, refrain 
to hit me hard on the face. I'm already mourning....

Many thanks in advance for your help !


* : The 3 disks are here because at one time, long ago, I used mirroring 
in the LVM software. ... I switched to software raid and kept the 3 
disks as I had them into the box.


-- 
"If a man empties his purse into his head, no man can take it away from him.
An investment in knowledge always pays the best interest"


Benjamin Franklin.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [linux-lvm] Sorry to ask here ...
  2017-02-23 19:25 [linux-lvm] Sorry to ask here Georges Giralt
@ 2017-02-23 20:23 ` Xen
  2017-02-24 17:00   ` Georges Giralt
  2017-03-02  5:44 ` Eric Ren
  1 sibling, 1 reply; 4+ messages in thread
From: Xen @ 2017-02-23 20:23 UTC (permalink / raw)
  To: linux-lvm

Georges Giralt schreef op 23-02-2017 20:25:

> Lastly, I bet the cause is an upgrade of some packages, the machine
> refuses to boot. It drops onto initramfs shell because the vg0 volume
> group is not activated. Launching lvm and doing a vgchange -a y does
> the trick and the system boots fine afterwards.

I don't know what is the cause, but...


The /boot is on the same partition so it should be no problem right.

In the initrd the root is opened but it fails?

The quickest thing to do is to go to:

/usr/share/initramfs-tools/scripts/local-top

And edit lvm2

You must add a call to "/sbin/lvm vgchange -ay" at the bottom. If this 
works it will try to activate everything while still being in the 
initrd.

Normally only root and swap are activated. I assume the mdX array will 
be assembled just fine, but I am not sure. However I think the chance is 
high that it would work.

There is a call in there I think to activate only $ROOT and $RESUME. 
Maybe it will work if you just vgchange -ay.

If you can enter the system in a chroot and edit this file, and then 
regenerate the initramfs, it will be included in the new initramfs.

If "lvm2" is upgraded this file will be overwritten.

I assume you can override it in 
/etc/initramfs-tools/scripts/local-top/lvm2

You could need to copy lvm2 there and edit it there, for that:

mkdir -p /etc/initramfs-tools/scripts/local-top

cp /usr/share/initramfs-tools/scripts/local-top/lvm2 
/etc/initramfs-tools/scripts/local-top

But I am not 100% sure of that.



In any case that would definitely add a call to "vgchange -ay" to the 
initramfs, so that the entire volume group can be activated prior to 
passing control to systemd.

Good luck.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [linux-lvm] Sorry to ask here ...
  2017-02-23 20:23 ` Xen
@ 2017-02-24 17:00   ` Georges Giralt
  0 siblings, 0 replies; 4+ messages in thread
From: Georges Giralt @ 2017-02-24 17:00 UTC (permalink / raw)
  To: linux-lvm

Thank you Xen,

Your trick is working fine.

As far as I can tell, the two md are properly working during boot.

Now, I've to analyze the boot and update logs to try to find what was 
the cause. But, the PC is booting fine. The wife won't complain. ;-) I'm 
happy ;-)

Thank you.


Le 23/02/2017 � 21:23, Xen a �crit :
> Georges Giralt schreef op 23-02-2017 20:25:
>
>> Lastly, I bet the cause is an upgrade of some packages, the machine
>> refuses to boot. It drops onto initramfs shell because the vg0 volume
>> group is not activated. Launching lvm and doing a vgchange -a y does
>> the trick and the system boots fine afterwards.
>
> I don't know what is the cause, but...
>
>
> The /boot is on the same partition so it should be no problem right.
>
> In the initrd the root is opened but it fails?
>
> The quickest thing to do is to go to:
>
> /usr/share/initramfs-tools/scripts/local-top
>
> And edit lvm2
>
> You must add a call to "/sbin/lvm vgchange -ay" at the bottom. If this 
> works it will try to activate everything while still being in the initrd.
>
> Normally only root and swap are activated. I assume the mdX array will 
> be assembled just fine, but I am not sure. However I think the chance 
> is high that it would work.
>
> There is a call in there I think to activate only $ROOT and $RESUME. 
> Maybe it will work if you just vgchange -ay.
>
> If you can enter the system in a chroot and edit this file, and then 
> regenerate the initramfs, it will be included in the new initramfs.
>
> If "lvm2" is upgraded this file will be overwritten.
>
> I assume you can override it in 
> /etc/initramfs-tools/scripts/local-top/lvm2
>
> You could need to copy lvm2 there and edit it there, for that:
>
> mkdir -p /etc/initramfs-tools/scripts/local-top
>
> cp /usr/share/initramfs-tools/scripts/local-top/lvm2 
> /etc/initramfs-tools/scripts/local-top
>
> But I am not 100% sure of that.
>
>
>
> In any case that would definitely add a call to "vgchange -ay" to the 
> initramfs, so that the entire volume group can be activated prior to 
> passing control to systemd.
>
> Good luck.
>
> _______________________________________________
> 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/
>

-- 
"If a man empties his purse into his head, no man can take it away from him.
An investment in knowledge always pays the best interest"


Benjamin Franklin.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [linux-lvm] Sorry to ask here ...
  2017-02-23 19:25 [linux-lvm] Sorry to ask here Georges Giralt
  2017-02-23 20:23 ` Xen
@ 2017-03-02  5:44 ` Eric Ren
  1 sibling, 0 replies; 4+ messages in thread
From: Eric Ren @ 2017-03-02  5:44 UTC (permalink / raw)
  To: LVM general discussion and development, georges.giralt

[-- Attachment #1: Type: text/plain, Size: 1937 bytes --]

Hi,

On 02/24/2017 03:25 AM, Georges Giralt wrote:
> Hello !
>
> I'm sorry to ask for help here, but I'm lost in the middle of nowhere ...
>
> The context :
>
> A PC hardware with an UEFI "BIOS" set to legacy boot only.
>
> 3 disks with 2 partitions each. The 3 first partitions are set in software mirror (md0) 
> and used for /boot (Ext4) .The second 3 partitions are used as a software mirror and the 
> resulting md1 is the sole PV of a vg0 group onto which 8 logical volumes are set. This is 
> used to run Ubuntu 16.04.2 software (after many upgrades...).   *
>
> This setup has worked for years and seen changes in disk sizes, main board, etc...
>
> Lastly, I bet the cause is an upgrade of some packages, the machine refuses to boot. It 
> drops onto initramfs shell because the vg0 volume group is not activated. Launching lvm 
> and doing a vgchange -a y does the trick and the system boots fine afterwards.
>
> I've searched everything I can (but maybe the setup was so rock solid I became lazy or 
> unconscious) and tried to make again the initramfs with no luck.
>
> So every leads or advice would be greatly appreciated. Please, refrain to hit me hard on 
> the face. I'm already mourning....
>
> Many thanks in advance for your help !
>
>
> * : The 3 disks are here because at one time, long ago, I used mirroring in the LVM 
> software. ... I switched to software raid and kept the 3 disks as I had them into the box.
Looks very similar with these bugs:

https://bugzilla.opensuse.org/show_bug.cgi?id=964862
https://bugzilla.opensuse.org/show_bug.cgi?id=919284
https://bugzilla.opensuse.org/show_bug.cgi?id=1019886

You might be interested in having a try with this fix (attached):
"simplify-special-case-for-md-in-69-dm-lvm-metadata.patch"

This patch was originally discussed here - https://www.spinics.net/lists/raid/msg55182.html

But for some reason, it is not accepted by upstream:)

Hope it can help.

Eric

>
>


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Simplify-special-case-for-md-in-69-dm-lvm-metadata.r.patch --]
[-- Type: text/x-patch; name="0001-Simplify-special-case-for-md-in-69-dm-lvm-metadata.r.patch", Size: 0 bytes --]



^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2017-03-02  5:46 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-02-23 19:25 [linux-lvm] Sorry to ask here Georges Giralt
2017-02-23 20:23 ` Xen
2017-02-24 17:00   ` Georges Giralt
2017-03-02  5:44 ` Eric Ren

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.