[Openstack-operators] Fwd: [openstack-ansible] We need to change!
jean-philippe at evrard.me
Tue Apr 17 15:04:07 UTC 2018
Sorry for the cross-posting, but I think this could interest operators too.
Starting at the end of this month, I won't be able to work full time
on OpenStack-Ansible anymore.
I want to highlight the following:
Our current way of working is not sustainable in the long run, as a lot
of work (and therefore pressure) is concentrated on a few individuals.
I managed to get more people working on some parts of our code
(becoming cores on specific areas of knowledge, like mbuil on
networking, mnaser and gokhan on telemetry, johnsom on octavia,
mugsie on designate), but at the same time we have lost a
core reviewer on all our code base (mhayden).
I like the fact we are still innovating with our own deployment tooling,
bringing more features in, changing the deployment models to be always
more stable, more user-friendly.
But new features aren't all. We need people actively looking at the
quality of existing deliverables. We need to stretch those
responsibilities to more people.
I would be very happy if some people using OpenStack-Ansible
would help on:
* Bugs. We are reaching an all-time high amount of bugs pending.
We need people actively cleaning those. We need someone to
organize a bug smash. We need people willing to lead the bug
triage process too.
* Releases. Our current release process is manual. People
interested by how releases are handled should step in there
(for example, what does in, and at what time).
We also need to coordinate with the releases
team, and improve our way to release.
* Jobs/state monitoring. I have spent an insane amount of time
cleaning up after other people. That cannot be done any longer.
If you're breaking a job, whether it's part of the
openstack-ansible gates or not, you should be fixing it.
Even if it's a non-voting job, or a periodic job.
I'd like everyone to monitor our zuul dashboard, and take
action based on that. When queens was close to release,
everything job was green on the zuul dashboard. I did an
experiment of 1 month without me fixing the upgrade jobs,
and guess what: ALL (or almost ALL) the upgrade jobs are
now broken. Please monitor  and actively help fixing
the jobs. Remember, if everyone works on this, it would
give a great feedback to new users, and it becomes a
* Reduce technical debt. We have so many variables, so many
remnants of the past. This cycle is planned to be a
cleanup. Let's simplify all of this, making sure the
deployment of openstack with openstack-ansible ends
up with a system KISS.
* Increasing voting test coverage. We need more code
paths tested and we need those code path preventing
bad patches to merge. It makes the reduction of
technical debt easier.
Really thank you for your understanding.
More information about the OpenStack-operators