openstack team mailing list archive

[Michael Still <mikal@xxxxxxxxxxx>]

Heya,

I would also like to run for the Nova PTL position. Let me start out
by saying that Russell has my complete confidence and would do a
fantastic job as a PTL. However, I think its a little unhealthy that
we keep having elections with only one candidate and I'd like to see
us develop a culture of healthy elections. I would say that I think
regardless of who wins I'd like to see more delegation in the Nova PTL
role, which will make the job easier and more enjoyable.

*** Who am I? ***

I am a senior software developer at Rackspace Hosting. Before that I
was a operations engineer at Canonical, and before that a Site
Reliability Engineer at Google. I've been working in "cloud" in some
form or other since before it had a name.

I have been working on Nova since late 2011, mostly focussed on the
libvirt hypervisor driver. I am also an Oslo core reviewer and serve
on the OpenStack Vulnerability Management Team. I am a frequent code
reviewer, and have contributed a significant body of code to Nova over
the last year. I've focussed mostly on Nova because I think its an
interesting set of problems, but also because I think its the way
compute in general will be managed in the future -- for example if we
all end up with racks of 3,000 ARM machines, I'd like us to live in a
world where that is managed by something like the baremetal
hypervisor.

I was also the Director for linux.conf.au 2013, which is the largest
event covering OpenStack to be held in Australia so far. I mention
this because I feel the experience is relevant to Nova in that the
conference is run with an entirely volunteer staff. There are strong
parallels between managing a hundred volunteers for a two year long
project and being Nova PTL and I think I've learnt some things about
running volunteer teams that can be applied to Nova. Specifically I'd
like to see more delegation occur. This doesn't have to be super
formal -- I'm thinking more along the lines of someone leading the bug
triage effort or ensuring code reviews happen in a timely manner.

I also have the support of my employer, who has said that they will
allow me the time required for this role, for which I thank them.

*** Platform ***

I think there are a few key things that the Nova team needs to focus
on in the next release or two. These are the things which will define
us as more people deploy OpenStack and have their first formative
experiences. We need to work hard to make sure that those experiences
are positive ones, because its much harder to fix a poor first
impression later.

* We need to be better about onboarding new developers to the project.
Our processes are complicated and somewhat poorly documented. To this
end I've proposed a session at the Portland OpenStack conference about
how to get up to speed as an OpenStack developer, but with a specific
focus on Nova. Regardless of what happens with this election or the
selection process for talks at the conference, I intend to vigorously
pursue helping new developers get up to speed on how to contribute to
Nova.

* We've done a great job of adding features of Nova, but now is the
time to burn down our bug backlog (over 500 bugs!), and get better a
triaging bugs in a timely manner. Similarly, we also need to be better
at getting a first review on new submissions -- its a pretty bad
experience when your first patch sits around for days waiting for its
first review. We can do much of that by providing more visibility into
how we're doing with those things so they don't get forgotten. I'd
like to see these issues consistently reported on at weekly Nova
meetings, and I'd like the community to start talking about more
complicated "who contributed to Nova" metrics than just lines of code.
I've done some early work on what some of these metrics would look
like, and you can see the results at
http://openstack.stillhq.com/reviews/report.html

* We need to take security more seriously. Often security
vulnerabilities sit for too long waiting for review from core. I will
acknowledge that I'm as much of the problem there as anyone, but we
need to focus on getting better at that. To that end, I'd also like to
see a more systematic effort to find security problems before they are
discovered by our users, and I'd like to see the Foundation help out
here with some security consulting if we can't find the resources we
need within the community.

* I think its clear that OpenStack is a great choice for public
clouds. We need to be working hard to make sure that our
implementation decisions work well for smaller private clouds as well.
OpenStack should ship with sensible defaults, and work well with
default configuration. Things are ok at the moment, but there is more
we can do to make this easier for our users -- for example HA is quite
hard to configure at the moment and can be made easier.

* Finally, I think the PTL needs to delegate more. The PTL doesn't
have to be trying to triage all bugs him or herself. They don't even
need to be driving the meetings for that matter. They should have a
trusted team of passionate assistants to help with the role. Its clear
we have team members who are passionate about different aspects of
making Nova great, and we should encourage that as much as possible.

*** Conclusion ***

To steal a little from Russell I'd also like to thank Vish for his
efforts. If it wasn't for his patient mentoring when I first joined
the project I probably would have walked away. Instead, I'm left with
a deep attachment to Nova and the community which runs it. I'd also
like to thank Russell for running. Its hard to put yourself out there
and run for something like this.