FAI dying during partition setup

[Ryan Steele]

Holger Parplies wrote:
> Hi,
>
> Ryan Steele wrote on 2008-09-29 17:13:29 -0400 [Re: FAI dying during partition setup]:
>   
>>>> Recently, I've been trying to set up LVM clients with FAI 3.2.4 for 
>>>> Ubuntu, using Peter Gervai's well-placed LVM hooks.  However, I keep 
>>>> hitting the situation where sfdisk cannot read the partition table, 
>>>> and mke2fs cannot create a filesystem, because it thinks the device 
>>>> is in use.  It's reproducible every time if I FAI-install the client 
>>>> more than once.  E.g., the first time I FAI-install the LVM-enabled 
>>>> host, it works.  [...]
>>>>         
>
> while I have no experience with FAI > 3.1.8 or the LVM hooks, I would guess
> that, for some reason, the volume group is activated (do you have anything in
> /dev/mapper?). 

I'll have to check.  I know that vgdisplay --verbose --partial had 
/dev/sda2 in it, but I couldn't vgremove it (again, got the BLKRRPART 
error).

> You could see if 'vgchange -an' changes anything about the
> BLKRRPART (that's one of the things I like about FAI - you can debug your
> scripts in the exact context they are run).
>   

I'll give that a shot as well.  And yeah, I like being able to drop to 
the emergency shell and run things right there - it's very helpful.

> I use FAI 3.1.8 with my own partition-hook for managing my own LVM requirements
> (like sometimes keeping some of the LVs ...). 

Completely home-brewed, or adapted from Peter Gervai/Michael Svamberg?

> What works for me without any
> problems (including repeated installation with identical disk layout) is
> basically
>
> - sfdisk
> - modprobe dm_mod
> - pvcreate -y -ff [and I didn't mention -ff ;-]
>   

Yeah, I changed Peter's script from --force to -ff due to some error 
messages.

> - vgscan
>
> and so on (including mke2fssing my root partition, which isn't on LVM). Of
> course I don't pvcreate in the cases I want to keep LVs :).
>
> I realize FAI 3.2.4 is a totally different context (kernel version, etc.), but
> maybe it helps just the same.
>   

Maybe, but before I start reinventing the wheel and writing my own LVM 
hooks, it might be worth my while to adapt 3.2.10 for Ubuntu.

>   
>>> 2. Added "dm_mod" to the list of kernelmodules in 20-hwdetect.source
>>>       
>
> So I guess my suggestion is, basically, to modprobe dm_mod later, specifically
> after the sfdisk has been done.
>   

Yeah, that might be worthwhile, because experience has taught me that if 
dm_mod is running via the initramfs, sfdisk -R always fails to have the 
kernel re-read the partition tables.

> Regards,
> Holger
>   

I appreciate the insight and advice.  I'll let you know if any of it 
works for me.