[PATCH 12/12] add a quick HOWTO for getting this branch working

Sam Vilain sam at vilain.net
Tue May 2 04:52:37 CEST 2006

Holger Levsen wrote:

>Thanks for publishing your patches, could you also publish a complete branch 
>somewhere? Maybe I missed where you stated against which fai version those 
>patches should be applied.

They were against the svn head.  I'm working on getting them on

>Also a changelog would be nice, atm you need to read all patches to see what 
>they're doing...
>Also, which is much more difficult to handle, some of your patches apply 
>against fai, others against vserver :-(

The ones I posted to this list were against fai, the one on the savannah
page against util-vserver.

>One specific question: whats the difference between a vserver and childen ? 
>Both are vserver-vms, but the latter is just copied ? (If thats the case, I 
>think your approach is too limited.)


  - VSHOST is the role that should get you a vserver kernel and
util-vserver and the ability to set up children
  - VSCHILD is the role that you get for being a vserver guest

When installing a VSHOST, all the linked VSCHILD guests should be
installed at the same time.

>I dont think your approach of modifying vserver-build is necessary, I just 
>build my vservers with vserver-build -m debootstrap and then exec into the 
>vserver and run fai softupdate. 

My major concern is making sure that a regular FAI build can be
prototyped using a vserver FAI build.

So, if you build the base system for one using task_extrbase and the
other using debootstrap, then you have lost some reproducibility. 
However it is probably an academic concern, given that the tarball is
just the result of a debootstrap anyway.

One thing this approach does let you do easily is FAI install 32 bit
vservers on a 64 bit host system, using a 32-bit vserver with fai
installed within.


More information about the linux-fai mailing list