On Fri, Mar 30, 2018 at 4:31 AM, Sam Doran <sdoran@redhat.com> wrote:
>> Ansible RPMs are already there
>> http://releases.ansible.com/ansible/rpm/release/epel-7-x86_64/ but they
>> depend on EPEL for additional deps.
>
> Ansible RPMs have always been there. I don't believe they depend on anything
> in EPEL.

You are correct, I had some stale info or mixed it up with something else.
Here is yum install output on an empty CentOS7 machine:
Installing:
 ansible              noarch 2.4.3.0-1.el7.ans /ansible-2.4.3.0-1.el7.ans.noarch
Installing for dependencies:
 PyYAML               x86_64 3.10-11.el7       base
 libyaml              x86_64 0.1.4-11.el7_0    base
 python-babel         noarch 0.9.6-8.el7       base
 python-cffi          x86_64 1.6.0-5.el7       base
 python-enum34        noarch 1.0.4-1.el7       base
 python-idna          noarch 2.4-1.el7         base
 python-ipaddress     noarch 1.0.16-2.el7      base
 python-jinja2        noarch 2.7.2-2.el7       base
 python-markupsafe    x86_64 0.11-10.el7       base
 python-paramiko      noarch 2.1.1-4.el7       extras
 python-ply           noarch 3.4-11.el7        base
 python-pycparser     noarch 2.14-1.el7        base
 python-setuptools    noarch 0.9.8-7.el7       base
 python2-cryptography x86_64 1.7.2-1.el7_4.1   updates
 python2-pyasn1       noarch 0.1.9-7.el7       base
 sshpass              x86_64 1.06-2.el7        extras

> sshpass and paramiko come from Extras, python2-cryptography comes from
> updates.

My concern is that if those were included in Extras for Ansible, they
would be removed from Extras together with ansible.

> I'm not sure if any of that is helpful since you mentioned it would need to
> be built by the appropriate SIG anyway.

Yes, ideally we would be able to get ConfigMgmt SIG going, in the
meantime other SIGs are rebuilding on their own e.g. Virt SIG/oVirt
did 2.4.3 http://cbs.centos.org/koji/buildinfo?buildID=21591
As a quickfix, we could also temporarily push this to RDO deps repo,
until we have rest of the plan ready.

>> BTW ideal approach would be to insert OpenStack use-cases into Ansible
>> upstream CI and make it voting, this could become reality with cross-project
>> CI efforts lead by openstack-infra. With that, Ansible master would never
>> break us!
>
> I don't entirely follow this, but I think it sounds like what I proposed
> above: having OpenStack test the devel branch of Ansible so Ansible
> Engineering can get feedback quickly if things are broken prior to a
> release. I know some of the OpenStack infra folks, and the networking team
> within Ansible has been doing a lot of work with them with Zuul for
> distributed CI. Myself and Ricardo Cruz on the Ansible side are very
> interested in hooking up more testing of Ansible as it relates to OpenStack
> using Zuul run by OpenStack Infra. Ricki and I talked about this a bunch at
> the PTG but have been working on other things since we got back.

Yes, above was forward-looking CD world where, given infinite CI
resources, everything is tested pre-commit across collaborating
projects.
Definitely trunk RPMs from devel branch are the step in that
direction, progression scale is:
no testing, push the latest release, hope for the best -> CI with
latest release -> CI with devel branch -> CI pre-commit

Cheers,
Alan