Server Admins: Why not Fedora?

Kevin Fenzi kevin at scrye.com
Thu Nov 7 20:50:12 UTC 2013


On Thu, 7 Nov 2013 21:22:44 +0100
Miloslav Trmač <mitr at volny.cz> wrote:

> On Wed, Nov 6, 2013 at 3:25 PM, Simo Sorce <simo at redhat.com> wrote:
> > I came into this thinking we wee going to build a Server oriented
> > distribution on top of the core, but since yesterday's meeting I am
> > not so sure that is the goal anymore, it seem people want to do
> > specialized sub-server 'products' (whatever that means, and I agree
> > with mizmo it's a confusing term to use).
> >
> > I really would like some more clarity of goals for each WG.
> 
> I think we basically "own the server space for Fedora", and how
> exactly we decide to cover it is up to us.  If the our plan is not
> clear, or there is a disagreement, let's keep talking and get it
> resolved.

+1

> From my point of view, the "specialized products" / "featured
> application stacks" view means that we want to achieve more
> inter-package integration and more common functionality, than a
> "server oriented distribution" typically does, and in particular more
> than Fedora has been doing.
> 
> And the talk about products seems to reflect an implicit consensus to
> get there one "application stack" at a time, rather than to cover all
> applications from the start, and add one distribution-wide feature at
> a time to all applications.

Yeah. 

I'm afraid I'm a technical/practica person so I can't help but keeping
thinking about this in terms of what we intend to integrate, produce,
test and distribute. 

There's a spectrum here. On one end we could keep things much as they
have been in the past: 

- We produce a netinstall iso
- We provide a repo with 10,000 'server' applications in it. 
- We test that the netinstall iso boots and works. 
- We try and address really critical bugs in really popular
  applications. 
- It's up to end users to make kickstarts and configure things for
  their needs. 

Pros:
- We already do much of this
- We don't have to make many decisions. 

Cons: 
- We can't really improve integration too much. 
- We don't help improve things much at all. 

All the way to: 

- We produce a netinstall iso. 
- We produce a dvd iso/image. 
- We produce curated comps groups for popular applications. 
- We provide a repo of 10,000 other applications on a 'best effort'
  basis. 
- We test that the netinstall iso boots and works. 
- We address bugs in our curated groups of applications. 
- We test our curated groups of applications and confirm they meet some
  set of test plans. 
- We produce some set of kickstarts or CM 'scripts' to help configure
  applications. 
- We produce some way to container applications so they don't interfere
  with each other. 
- We setup a common auth/CM/container/management/montitoring solution. 

Pros:
- Users would know that an application group was tested/works. 
- Users could base ks or CM stuff on things we produce. 
- Users could have something working, doing a lot of stuff out of the
  box. 

Cons: 
- we have to decide what curated applications we wish to make. 
- We have to make test plans. 
- We have to test things for release and commit resources to fix
  blockers. 
- We have to write integration between things that aren't integrated. 
- We have to make difficult decisions on what is our 'solution' in
  spaces where there are multiple solutions. 

I think we need to look at some in between state for short term, and
investigate how we get to the more difficult things longer term (or
even if they are practical). 

kevin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/server/attachments/20131107/3aba844b/attachment.sig>


More information about the server mailing list