On Tue, Jun 3, 2014 at 8:55 AM, Colin Walters <walters(a)verbum.org> wrote:
On Mon, Jun 2, 2014, at 12:43 PM, Simo Sorce wrote:
> The 2 roles we have in Fedora Server for now are FreeIPA and Database
> server. The 'meat' for both of them is in the data they provide, so I
> see little value in trying to use containers for them.
I'd say again there are a collection of factors here, and then agree
you that those factors do weigh in favor of traditional deployment.
> Although I can
> totally see how we may want to containerize them as an option in F22 as
> a way to upgrade the host but keep running the old code.
Yes, though that raises the question about the support lifecycle of
the dependencies. If you're talking about shipping F21 FreeIPA
and rebasing the host to F22 (or equivalent for downstreams),
I would say that sort of thing is potentially a major win for
both the upstream and the downstream.
We could without changes in the upstream (Fedora in this case)
lifecycle allow running F21 containers on F22, as the component RPMs
are supported for N and N-1.
In practice with Fedora today though, the kernel for example is already
in lockstep between N and N-1 and has been for some time - encouraging
old apps on new host would just be a more flexible version of the
Just to clarify, it's not quite in lockstep. We rebase on N, wait for
some semblance of actual testing and positive karma, and push that to
stable. We wait a bit longer after that to see what the fallout is,
and only then do we start thinking about rebasing N-1. There's often
a 2-3 week lag before N-1 is rebased, with subsequent builds of
whatever kernel N-1 is already on being done in the interim.
Also, we have an exception from FESCo for all of this. (Not to be
overly paranoid, but people shout at us constantly for not following
the Fedora update rules and I'm tired of repeating myself.)