Greetings.
I've been meaning to send this out for a while, but keep getting
sidetracked, so today I decided to sit down and get it sent. ;)
Right now, we have a process for new things that come into Fedora
Infrastructure called Request For Resources. This was a process started
long long ago, and revamped a few years ago. The idea of this process
was that it would involve existing infrastructure folks (you MUST get a
sponsor from sysadmin-main for any new RFR), that it would ensure
things like enough people know how the service works that we can
maintain it, that it's monitored, etc. Ie, supported by Fedora
Infrastructure.
This has I think worked pretty well in the past, but we are running
into a number of things in the last few years that this process just
doesn't work right with.
I don't want to suggest yet any solution to this beyond adjusting our
RFR process, but I will start the discussion with some questions:
* In the past we pretty much had a binary support status: supported or
not supported. We kind of added a middle area with jenkins:
supported, but you shouldn't depend on it, we will fix it when we
like. Should we have different support levels? How can we set users
expectation for what level the thing they care about is in?
* Should there be some requirements/expectations for a thing having a
fedoraproject.org dns entry? ie, there have been people who have
approched us for that for applications that they run, etc. How can we
explain to users when something like that is down who to contact?
* What level of support should we say cloud instances have?
* Adding some new service/application to an existing application,
should that follow the entire RFR process? or not since its just
adding to an existing application?
Any other thoughts or questions around this? I look forward to hearing
from folks. ;)
kevin