On Wed, 16 Oct 2013 05:27:07 -0400 (EDT)
Kamil Paral <kparal(a)redhat.com> wrote:
> > With the conversation we've been having and the persona
> > developments, I propose the following:
> >
> > - create FAS groups for git-taskbot-core and git-taskbot-tasks
> > which will control access to the git repos
>
> What will git-taskbot-tasks be used to? I thought we were going to
> have decentralized system of git repositories with test cases. Is
> this going to contain just some core test cases provided by Fedora
> QA?
Yeah, mostly some core test cases - the idea is still to have
decentralized git repos to store tasks. Would a change in name make
that more clear?
Rather than name change, I assumed we would host our Fedora QA tasks in a Fedora
QA-owned&branded git repo. Something like git-qa-taskbot-tasks would be one of the
sources that taskbot could pull. But I don't mind having our tasks closer to the core
repo, that's fine.
Which reminds me, will we be trying to have Taskbot at least partially distribution
independent, for example something that OpenSUSE and other RPM-based (or maybe not just
RPM-based) distros could pick up as well? It would add more abstraction layers (e.g.
abstract out fedmsg). But we could also gain external contributions. I don't think
we've pondered about this too much, have we?
This also reminds me of an investigation to do - figure out how hard a
git server would be hit if we stored task repos on it. If a repo is
cloned every time that a task is run, that's a lot of clones. I know
that rhodecode had a hard time not crashing in that situation but
gitolite + cgit seem to be doing fine.
A logical step would be to clone the repos on the server (i.e. simply check up-to-date
status on every task run, or regularly) and copy the files onto the test client during
client initialization. We will need to support file transfer (probably bidirectional)
anyway.