On Fri, 26 Jul 2013 12:23:33 -0600
Tim Flink <tflink(a)redhat.com> wrote:
After looking at the jobstats on autoqa-stg over the last day or so
since I changed all of the clients to be fc18, I'm happy enough with
the completion rates to suggest that we upgrade production autoqa and
get rid of our fc17 clients.
Proposed setup:
2x virthosts with 4x VMs each, 4096M and 20G disk
- 6x f18 x86_64
- 2x f18 i386
This is the setup I've been using in stg for a while and it seems to
be able to keep up with incoming jobs just fine. If it turns out not
to be enough, we can assign another virthost to production. With the
new ansible playbooks, it's trivial to reconfigure/add/remove clients
and the full process from virthost install/config to client
install/config would take well under an hour.
I'd like to do this on either Tuesday or Wednesday when we're removing
the fc17 repos from autoqa production. At that time, I'll upgrade
autoqa01 with the test fixes from the last several days, remove the
fc17 repos and disable the conflicts test as a non-hotfix.
Are there any objections to this plan or any additional details
requested before making a decision?
Tim
There have been no objections to my proposed plan and I'm taking that
as a "ok, go ahead and do it".
I'll be pushing fixes to git today and I'm planning to push a new
autoqa release to our epel testing repo so that it can run for a while
before moving to production. I'll take down production autoqa tomorrow
around 20:00 UTC for updates and reconfigure/rebuild the clients and
virthosts.
As part of this, I'll be removing the fc17 repos and removing the
conflicts test from scheduling as it continues to have issues on fc18
and as far as I know, nobody is looking at the output.
If you have any objections to my plan, speak now or forever hold your
peace :)
Tim