On Mon, Apr 04, 2016 at 02:45:51PM -0600, Kevin Fenzi wrote:
On Mon, 4 Apr 2016 22:06:46 +0200
Michael Scherer <misc(a)zarb.org> wrote:
> On Mon, Apr 04, 2016 at 01:26:28PM -0600, Kevin Fenzi wrote:
> > This one I am not sure about how best to handle.
> >
> > If we simply apply this and run it we have the modified service
> > files still around. Perhaps we could do this in 2 steps? Replace
> > the service file with the one from the package, then after
> > everyplace is updated, remove that?
>
> > Or I suppose we could reinstall the package, but that might cause
> > other issues.
> >
> > kevin
>
> IIRC, the config in /usr is less prioritary than the one in /etc, so
> keeping the old one in /usr should not cause any issues, and will
> solve itself with next package upgrade or reinstallation of the
> server.
>
> Since that's for fedmsg-gateway, i suspect that we will one day
> commit code to it and upgrade anyway.
Well, I don't like the idea of the machines being in a state thats not
described by the playbooks. Sure, it will go back on upgrade, just rubs
me the wrong way. ;)
So, since the fix is just yum reinstall of the package, we can either do that
with a adhoc run, a playbook or adding code to the role. I am not keen on adding
code to git just for a 1 off task and remove it just after, but if you think that's
better, i can add it.
--
Michael Scherer