[arch-general] [arch-dev-public] [signoff] device-mapper 1.02.29-1 & lvm2 2.02.43-1

[Tim Gelter]

RedShift wrote:
> Tim Gelter wrote:
>> Thomas Bächler wrote:
>>> Dieter Plaetinck schrieb:
>>>> I'm a user, not a dev, running on i686.
>>>> I couldn't find a definition of a 'signoff', but I updated abs, built
>>>> the 2 new packages, they compiled fine, i installed them, rebooted my
>>>> system and everything came up fine (dm_crypt+lvm based system).
>>>> I also tested the basic commands (pvdisplay,lvdisplay) and even did an
>>>> lvextend and resize2fs on a volume. everything still works fine.
>>>> (didn't
>>>> try making new/deleting LV/PV's etc though)
>>>> Does this count as a "signoff" ?
>>> The point of the signoff is not that you can build the package, but that
>>> the package provided in the repositories is working. If you can build
>>> the package, you still won't know if (for example) one of the
>>> executables inside the binary package is corrupted. Other than that,
>>> your tests are sufficient.
>>>
>>>> PS: Users cannot send to arch-dev-public at archlinux.org, so fwiw I cc'd
>>>> arch-general.
>>> Indeed they can't, you can always reply on arch-general if you feel the
>>> need to comment on a developer discussion.
>>>
>> I believe that all answers to this question have been given so hopefully
>> you don't mind if I hijack this thread...
>>
>> Since we're in the middle of discussing LVM, I've got a request.
>> Not too long ago, I had my root partition "/" on an LVM physical volume.
>> It was actually pretty easy to set up and worked like a charm until I
>> created a snapshot of one of my other PVs. As soon as I did so, I could
>> no longer boot the system because the dm_snapshot kernel module was not
>> loaded. I was told by several people that I needed to add a "hook" for
>> it, but never did figure out how to do so.
>> It seems to me that if we are going to support booting from logical
>> volumes, we also need to make sure that users who do so are still able
>> to boot if they decide to take a snapshot of any logical volumes,
>> including booting from a snapshot of "/".
>> I'll be glad to help test and even work on the hook if anyone gives me
>> instructions detailing how to do so.
>> Thanks!
>> -Tim
>>
> 
> Put dm_snapshot in the MODULES array in /etc/mkinitcpio.conf and then
> rebuild the initial ramdisk using mkinitcpio -p kernel26 (assuming you
> are using the kernel26 package from core).
> 
> Glenn
Thanks for the recommendation Glenn. If I remember correctly, I did
exactly as you suggest but it didn't solve the issue. I think it was a
matter of loading the driver at the correct time during the boot
sequence and that was the reason for the HOOK. I actually was given some
vague instructions about what I needed to do from a dev in #archlinux
but never got around to doing it because I got busy with work and home
life. I currently don't have the setup to verify what needs to be done
which is why I posted to the list. I may of course be mistaken  and
won't be upset by any corrections I receive. :)
Thanks again.
-Tim

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 489 bytes
Desc: OpenPGP digital signature
URL: <http://archlinux.org/pipermail/arch-general/attachments/20081116/1fef4c9c/attachment.pgp>