7:40 a.m.
----- Original Message -----
From: "Tim Flink" <tflink(a)redhat.com>
To: "AutoQA development" <autoqa-devel(a)lists.fedorahosted.org>
Sent: Tuesday, November 13, 2012 6:46:54 PM
Subject: Current State of Image Building Project
I figure that one good place to start is to describe the current
state
of the project - what it can do and what the gotchas are. Amit, if
I'm
missing anything, please correct me or add what I'm missing.
The local part of image building works fine - there are some quirks
to
the process and a configuration file format to learn but the added
builds are downloaded and included in images. I haven't done much
testing with building live images but I believe that is also working.
All valid points, Tim. I just tested it today on a fresh Fedora 17 system and the
documentation is quite true to what you need to do. I haven't tried building live
images today, but all was working when I last tested it. One point to note in case of the
live image building is that, I am using the 'livecd-creator', rather than the
'livemedia-creator'. I had some problems back then (four months back) with the
latter, and hence used the former. Perhaps, at some point of time, we will need to look
into using the latter. The local image building should be simple to setup. The fabric
script has a task defined explicitly for that. Please see [1].
[1]
https://fedorahosted.org/releases/f/e/fedora-build-service/devel/doc/HOWT...
In order to use any of the building code, you need to have the
following installed:
koji lorax livecd-tools pungi pykickstart spin-kickstarts
Depending on which mode you are deploying the service, the dependencies are installed
accordingly. The fabric script takes care of installing the dependencies.
Virtualenv doesn't work unless you create it as root because the CLIs
need to be run as root for pungi/lorax and that changes paths to the
point that not all the code works. I'm sure there are ways to work
around it other than creating the virtualenv as root but I haven't
tried very hard to find one.
I haven't tried running with Virtualenv, so can't comment.
The remote parts work once set up - I had it set up at one point
without the web front end and was able to build images. It needs a
bit
of polish but the needed stuff is mostly there, I just found the
setup
process to be a little touchy and fragile. The current method of
configuration is rather pre-defined and not incredibly flexible but
the
basic method is sound.
Completely agree. I am pretty sure you can get a "working" deployment. But,
there are rough edges which need ironing. Hopefully, some of the current tickets, once
implemented will allow us that.
The web interface wasn't a priority during GSoC and thus, it is
pretty
rough around the edges. I was never able to get the full stack
working
but it's certainly capable of working. The UX needs more designing
and
a re-skin but this stuff can be done at a later date.
+1.
I think that sums it up relatively well. The docs should fill in any
more details that I've not written enough to start using the code. If
not, feel free to ask questions here or in #fedora-qa.
Yes, I am positive about the documentation being a good place to start. It should give you
an idea of what is currently possible with it.
Best,
Amit.