Upgrading or Replacing AutoQA

[Tim Flink]

Apparently, I'm not paying enough attention to the F20 scheduling
process because the branch date is coming up far faster than I ever
expected (2013-08-06 is the currently listed date). I realize that
isn't set in stone yet but I also think it would be naive to plan on
that date moving back.

AutoQA is currently running on a mix of F17 and F16 clients (mostly
F17, there are only a couple of clients still running on F16) and with
F17 going EOL soon, my hope was to get taskbot up and running soon
enough to depreciate AutoQA before running on F17 was an issue and we
could avoid having to rebuild the AutoQA infrastructure.

With the schedule as it is now, I think that planning for taskbot to be
ready in time for F20 branch is too risky and pretty much impossible in
less than a month. I'm proposing that we take the time to rework the
AutoQA infrastructure so that it is running on a mix of F18 and F19
(maybe some rawhide) instead of leaving it as is.

If we rework the infra, I want to ansible-ize everything. The current
by-hand method of admin is crazy and way too much work to re-deploy or
even run updates on. I also want to tweak the db config a bit in the
hopes that we can get a bit better performance out of autotest in
production.

I'm going to discuss the possible changes with infra and put together a
more detailed proposal but wanted to see if there were any objections
or alternative proposals out there. If you're interested, more help
would also be appreciated :)

Tim
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/qa-devel/attachments/20130708/478eea2a/attachment.sig>