On Mon, Jul 30, 2012 at 11:55 AM, Ralf Schmitt <ralf(a)brainbot.com> wrote:
It's available on the any of the pdf cluster
machines in /home/pp/local
(together with the software, which was a mistake).
Creating a puppet configuration and debianizing the pediapress software
according to the ops teams requirements is an (ongoing?) project. I did
the latter and if you feel like having the energy to push the former
forward please do so.
Great. I didn't know about any of that. I don't have access to those
boxes so I can't just pull it off the machines directly. Can it be
re finding someone to puppetize: I can't commit to anything for at
least a week; will reevaluate then unless someone else has done it
I've been administrating the pediapress software
installation on these
machines without receiving any gratitude. In fact I received quite the
Sorry! I asked around (before my original message) and no one seemed
to know where the config was stored so I mailed the list. Actually I
didn't even know you were the administrator until reading that last
Asking me to "commit to maintain it" just
feels wrong to me. The ops
team wants to control these machines, so they should do it!
I think it's perfectly reasonable. Config for the cluster should be
stored in gerrit and that should (ideally) be the canonical copy; at
any rate, changes to the cluster should never propagate to gerrit as
an afterthought, rather they should be propagated as an early step in
the process of making a change to the cluster. Getting a snapshot of
one point in time that is then never updated is useful (for a little
while) but having a copy that's always in sync with what's actually
running is also quite important.
Note: This doesn't need to be puppet even. (although it would be
eventually moved there naturally I guess) It could just be putting
some # of sub dirs of /etc into git. (or /home/pp/local ? idk the
layout...) Again, I'm volunteering to sanitize for you and if WMF ops
already has access to the path you named then that may be good enough.