LTS Backports Maverick EOL

[Andy Whitcroft]

On Wed, Apr 11, 2012 at 02:25:08PM +0100, Andy Whitcroft wrote:
> On Wed, Apr 11, 2012 at 07:06:06AM -0600, Tim Gardner wrote:
> > On 04/11/2012 06:28 AM, Andy Whitcroft wrote:
> > > As Maverick is now officially EOL we will no longer be producing further
> > > kernel updates for that release.  This wil orphan any consumers of the
> > > LTS Backport Maverick kernel for Lucid.  The security team is recommending
> > > we migrate them to a later LTS backport kernel.  I tend to agree it does
> > > not seem appropriate to just leave them silently with no further updates.
> > > This leaves the question as to where they should be migrated to.  Any jump
> > > to a later kernel is going to be risky.
> > > 
> > > If we are going to jump them to a later LTS backport with the inherant
> > > risks it likely makes sense to jump them to the latest LTS backport that
> > > exists for their release to gain them the maximum support and prevent
> > > us having to do the same again in six months when the following backport
> > > drops from support too.
> > > 
> > > I have a patch in my tree to jump them forward assuming we agree on
> > > where they should go.  I propose Oneiric.
> > > 
> > > Comments?
> > > 
> > > -apw
> > > 
> > 
> > I think automatically updating to another kernel release is likely more
> > dangerous then not having security patches applied.
> > 
> > What about changing the Maverick kernel meta package reference to point
> > at a script that asks the user what they wanna do ?
> 
> As in a preinstall script to ask where they want to go
> 
>  o Stay here with NO security updates
>  o Move to Natty Backport
>  o Move to Oneiric Backport
> 
> I guess that way they at least find out about it in a forceful way.  I
> will talk to the gurus to find out if that is a viable way forward.

Talked this round with those in the know, and the feeling was this would
be very very complex packaging to get this effect and very hard to test
as we would only be giving it to a very small number of users relativly;
and we would not be able to test it in precise even before shipping it.

Overally there seems to be pressure for strong documentation in the
changelog as active server admins should be looking at those and
reacting to their contents.  Desktop combinations are not supported so
we should not have to worry about the rank and file hitting this.

It was also brought up that -security thought that this was always the
plan, that people would be upgraded through to the latest version.

-apw