RepoFunnel and the Software Component Pipeline

Colin Walters walters at verbum.org
Mon Aug 24 21:35:51 UTC 2015


On Sun, Aug 23, 2015, at 10:42 PM, Nick Coghlan wrote:

> Configuration management is only an effective approach for centrally
> controlled systems, 

I've been pretty happy using ansible so far for control over my laptop,
and a few machines, it's really not hard to do.

And definitely if you're talking about "multiple servers" is where
config/state management comes into play anyways, beyond
just delivery.

> I want my systems to be as close as possible to the
> default state in order to maximise the chances of trouble-free
> upgrades.

That statement also kind of clashes with installing a lot of COPRs =)

> Centralised configuration management also doesn't work for easily
> defining and sharing project specific repos containing pre-integrated
> package sets (e.g. the beaker-project.org repos, which provide not
> only Beaker, but also any dependencies which aren't part of the
> underlying operating system).

I'm not sure exactly what problem you're referencing here, but
I wasn't disagreeing that there is some value in more intelligent infrastructure
atop a set of rpm-md repos, whether they be COPRs or whatever.

> Filtered repo definitions, by contrast, are very easy to federate
> across different users, and very easy to consume when building
> container and machine images (regardless of the image creator's
> preferred choice of configuration management tool).

Maybe...if what you're saying is "installing 1-3 files in /etc/yum.repos.d
is easier than 5-10", possibly, although I myself haven't seen large
numbers there.



More information about the env-and-stacks mailing list