How do companies do OpenStack?

I’m cruising Silicon Valley these days asking “how does your company do OpenStack?”  to collect best practices and notable mistakes from various leaders of OpenStack corporate community. I’m hoping to build a ‘how to’ manual to help managers build better dev teams, more effective at collaborating while shipping products to their customers. This is an effort that goes hand-in-hand with training new developers with Upstream Training and other initiatives aimed at sustaining OpenStack growth.

I often have to explain to managers of OpenStack corporate members how and why to contribute to the project. The level of complexity reached has now exceeded the simple answers, like ‘fix simple bugs’ or ‘read the wiki pages’. For example, the participants to the Upstream Training in Atlanta had difficulties finding low-hanging-fruit bugs to work on: by the time they were ready to commit, someone else had already fixed it. Too much is going on at the same time, newcomers are confused at many different levels.

While we will keep training developers through Upstream Training (and other initiatives), we’re going to start talking to managers too.  I’d like to know how development teams are organized, how they balance shipping products to customers with contributing upstream, what incentives are in place that help or prevent contributions to go upstream, how career paths are influenced by contributions to the open source collaboration, etc.

If you’re managing a team of developers or devops, you’re a CTO, a HR executive or are managing in any way a team of engineers contributing code to OpenStack, I would like to talk to you: email me and vote for my talk proposal “How to maximize effectiveness of your developers contributing to OpenStack”.

Only developers should file specifications and blueprints

If you try to solve a problem with the wrong tool you’re likely going to have a frustrating experience. OpenStack developers have been using the same workflow to plan and manage development of OpenStack software over time, they chose a set of tools appropriate for the project’s software development. Developers use blueprints define the roadmap for the various projects, the specifications attached to a blueprint are used to discuss the implementation details before code is submitted for review.

These are the tools for the developers and this doesn’t mean that blueprints and specifications are the only way to interact with developers. Operators and users in general don’t need to dive in the details of how OpenStack developers organize their work and definitely should never be asked to use tools designed for and by developers. When I read ‘s post the place of a non-developer in the openstack community I immediately felt her pain: she was asked to use the wrong tool to solve a problem. I think this case is a major misunderstanding but the comments on the post signal that this is not an isolated case.

The most common way for a non-developer to highlight a flaw in OpenStack is to file a new bug report. Bugs can be defects that need to be fixed  or they can be requests for enhancement. In this case, Dafna filed a bug report and interacting with the triager, they agreed that the reported bug was not a defect per se but more of a request for enhancement, a wishlist.

In order to fix a defect or implement a wishlist item, developers need to file a blueprint (and a specification) before they can start writing code. If the person who filed the original bug report is also a developer then s/he can go ahead and continue carrying on the process, file a blueprint to solve a bug # and write specifications (when needed) to describe the details of the implementation. Users interested in the bug can chime in adding comments to the bug and to the specs, provide input to developers.

The process above works in similar way when the person filing the bug is not a developer, like in Dafna’s case. The proper flow of bug #1323578 would have not required Dafna to file a blueprint and specs, but to have a developer do that. Users are required to interact closely with the developers to discuss the implementation details, and that’s where the new specifications process helps. Gerrit may not have the best UI but it’s definitely better than holding discussions on a mailing list. Adding comments and holding conversations with the developer assigned to resolve the issue, on the bug report itself is also a valid option.

While I think that as a community we have plenty of ways to improve how we include users and operators in our conversations, in this particular case, I think frustration came from being pulled in the wrong place from the beginning.

PS: To be super-clear, in this post I’m using the terms developer and operator to describe a role, not a person. One person can be at the same time a developer and an operator, and act at times as a developer writing blueprints, submitting code and as an operator filing bug, giving comments to specifications.

 

Tracking gender diversity in the OpenStack developer community

The OpenStack Foundation has always tried to increase genders diversity in our community: we joined the Outreach Program for Women, established clear policies for our summits and in general we’ve been actively promoting good behavior across the board. A few weeks ago I realized that we were lacking of a decent way to track our progress in gender diversity in the OpenStack developer community and decided to improve the situation.

When Anne asked me for the number of women or non-male developers in OpenStack, I realized that we don’t have that number. When people register to become OpenStack developers they need to become members of OpenStack Foundation and have an account on Launchpad. Neither of them track gender information. The best number I had to offer was the t-shirt kind requested at our Summits. If you don’t measure it you can’t improve it, the saying goes. So we thought we need to offer the option for Foundation’s members to tell us their gender, if they want to. How do we ask for gender without being disrepectful to non-binary genders?

We did some research, discussed them on the bug and adapted the best practices to our case. We wanted to keep the choice short and not to offend anybody by leaving out options. An open entry form leads to hard to use data (people have very creative ways to spell just about anything), using pronouns would have made translations a nightmare. We need a way to restrict options so we can easily count them in the database so the binary options male/female were set first and ‘Prefer not to say’ was soon added as another option, since it’s really not mandatory to disclose that information. For non-binary genders using “Non-binary” sounded too geeky to me; using ‘other’ sounds weird, borderline offensive. We came to a consensus on my suggestion to have an open entry prefixed by “Let me tell you”. I liked this phrase because I feel like “Let me tell you” empowers the member to own their own gender definition. The new form is now live so you can now register or edit your OpenStack Member profile and add your gender (if you want to).

Why people contribute to OpenStack

A few days ago, I asked 55 OpenStack developers why they decided to submit one patch to OpenStack and what prevented them from contributing more. The sample polled people who contributed only once in the past 12 months, looking for anecdotal evidence for what we can do to improve the life of the occasional contributor. To me, occasional contributors are as important as the core contributors to sustain the growth of OpenStack in the medium/long term. Since it’s possible that the community has already already hired the best candidates among Python developers and system operators, to sustain the growth of OpenStack we need to tap into the wider Python community and experts of large distributed systems coming from other languages (Java, for starters). Knowing what hurdles new contributors face when they first join OpenStack is a first step to make onboarding easier and multiple contributions more frequent.

People who replied are for the most parts users of OpenStack, that is developers and dev-ops people working on deployments of OpenStack for internal usage and system integrators who develop and deploy OpenStack for their customers. A third group of occasional contributors are developers of projects that use or are used by OpenStack, like Docker or Ceph.

The top reasons to submit a patch went from the practical convenience of having patches maintained upstream, to fixing bugs (discovered during a proof of concept or in more advanced phases of a deployment) or to add/fix funcionalities to better integrate OpenStack with other projects. Finally, people contribute patches for personal growth and to enrich resumes.

Why don’t they submit more? There are three main stumbling blocks. Legal hurdles delayed contributions, and even prevented (at least) one. The reasons for the delays range from NDAs with customers to incompatibility with corporate policies to release software under open source license and sign a Corporate Contributor License Agreement. Another group of reasons for the delay is the long time it takes for contributions to travel the review queue. The time it takes for patches to get in OpenStack repositories was mentioned more than once. Finally, the lack of simple issues to fix in the spare time: apparently the community fixes simple bugs too rapidly.

The results cannot be considered representative but they seem to confirm that legal hurdles and speed of the reviews are preventing more casual contributions. While there are already many discussions on how to improve reviewing times, changing the way we handle contributions legally requires a massive endeavour to change OpenStack Foundation’s bylaws. The fact that simple issues to fix are hard to find is new to me: it may indicate that there are lots of people joining the community fixing these ‘low hanging fruits’ or something else. I think more analysis is needed in this area.

Exploring why people contribute to OpenStack

No doubt that OpenStack is one of the largest open source development efforts existing at the moment, with Linux, Debian and few others.  What can be improved to make the OpenStack community grow faster? How can we help new contributors become productive more rapidly?

Whatever dimension you look at, the growth of the OpenStack developers community is astonishing. I’ve looked at the reasons for this growth and I’ve identified four main causes summarized in my presentation at Linux Conf North America and Open World Forum in Paris last year:

  • One clear mission to solve a hard problem
  • One initial core team with strong motivation to solve that problem
  • One team dedicated to recruiting new team members
  • One clear path to onboard new members in the team

The first two points are pretty clear already and don’t seem to need much tweaking: those ain’t broken, so no need to fix’em. The other two are not broken either but they’re the kind of stuff you want to stay on top of because once you notice they’re broken, the cost to fix them will be higher. Recently I started to ask myself what margins of improvements there are in recruiting and onboarding new members in our communities.

I quickly realized that while we know a lot about the top contributors to OpenStack, we don’t know much about the ‘long tail’ of contributors. It’s fairly easy to understand why the people responsible for 80% of OpenStack code and documentation work on the project. But who are the occasional contributors and what are their motivations to join the community? I embarked on a quest to get to know this long tail in an effort to understand how we can improve recruiting and onboarding new developers.

After a conversation with Asheesh Laroia, of OpenHatch fame, I selected 55 random developers from the list of those who submitted a single change to the integrated OpenStack programs during the past 12 months. These people went through quite a huge chunk of trouble only to submit one single patch: besides learning enough of the quite complex OpenStack code, they had to sign the Individual CLA (maybe the Corporate CLA, too), learn about git review and the other tools, go through the code review process. I would expect that whoever goes through that process, does it with the intention to stick around. So I asked the sample two simple questions:

  • What made you decide to submit a patch to OpenStack?
  • What prevented you to submit more patches?

I’ll wait a few days for their answers and share more details. Feel free to answer the questions leaving a comment here, too.

How not to use OpenStack Community Metrics tools

I have noticed this morning another article/blog post mistakenly trying to extrapolate hard facts about a company’s involvement in OpenStack using one of the reporting tools built for the community. The reporter went to Stackalytics (but it could have gone to Activity Board, it would have been the same) to check if Oracle had made any contribution to the OpenStack codebase. That’s the wrong way to use these tools: numbers regarding companies contributing to OpenStack published daily cannot be trusted.

Both Stackalytics and Activity Board depend on data that is entered voluntarily by the contributors to OpenStack, therefore they cannot be trusted. Stackalytics has a mapping file in its repository that is kept up to date by developers themselves (those that know its existence). Activity Board pulls data straight from the OpenStack Foundation Members database: when you sign up as Member of the Foundation (a precondition to become a developer) you’re asked to enter data about who pays for your contribution. The bylaws of the Foundation also require that you keep that info up to date, but we know as a fact that few people log back in their member profile and even fewer update their affiliation. Therefore we know that the data about the affiliation in all reporting tools is not 100% reliable at any point in time. It’s good enough if you’re looking at the top contributing companies where the volumes are high enough to remain fairly valid despite small percentage errors. But when a reporter goes to check if a total newcomer to the community has submitted any code, that number is very likely to be wrong (and close to zero): the new developers may have not understood what the Affiliation field is and not filled it out (I see a lot of those on a weekly basis) and they’re very unlikely to know about the mapping file in git for Stackalytics.

The data that I trust most (but still not 100%, especially for ‘long tail’ contributors) are the reports published with Bitergia at release time: every OpenStack release we do a lot of manual cleanup of the data in the Foundation database, ask people to update their affiliation, normalize names of companies and circulate the report for comments before making them public. Still, those may contain errors which we track on Launchpad.

As far as I know the reporter didn’t ask the Foundation nor Oracle if anybody could point at actual commits done by Oracle employees and that’s what he should have done.

OpenStack prouds itself for being an open community and I’ve been the first proposer of having a public way to see the various activities inside the project, in real time and including the information about companies, not just individuals. I think we need to discuss how we can provide better data and avoid giving false illusion of precision to casual visitors to these sites.

How to get new questions on Ask OpenStack via email

Email is still one of the best mechanism to get push notifications, so getting automatic nudges via email from Ask OpenStack is useful. I took some time to document how to get an email message every time a new question is asked on Ask OpenStack.

The core of the configuration is available on the personal profile page. In my case the URL is https://ask.openstack.org/en/users/9/smaffulli/?sort=email_subscriptions: hit your username in the top navigation bar and then pick ‘subscriptions‘. The page will look like the screenshot below.

Selection_001
Askbot email settings

As of today Ask OpenStack gets around 10 new questions per day and the rate is increasing steadily. I highlighted the configuration that can help you prevent getting too many emails: select the frequency of the notifications want to get notifications first in the line “Entire forum (tag filtered)”. Then set the tag filter: I picked “only subscribed tags”. You can modify the tags you want to receive notifications from the form at the bottom of the page:

Selection_004
Add tags you’re interested in to this form

Tags are added by the person asking the question and can be edited by anybody with karma higher than 100. On the same subscriptions page there is also a setting to receive notifications for questions asked in Chinese languages.

On Ask OpenStack home page there are also ways to customize the site further, hiding questions carrying ignored tags from the view or highlighting interesting ones

Selection_002
Further configuration options on Ask OpenStack home page

 

Wrap-up post OpenStack Summit in Hong Kong

Back from busy days in one of the most exciting cities I’ve ever visited, I needed some time to put thougths back in order, recover from jet lag and deal with my irremediably broken WordPress installation (will probably blog about this later, too). Hong Kong was a blast in many aspects. The Summit itself started with lions dancing at the sound of drums in front of over 3,000 people:

And then we saw that Bejing is the first city by total number of contributors to OpenStack in the whole world:

There has been an incredible growth inside and around OpenStack, the project is growing fast. Growth is good, it’s what we wanted so we have plenty of reasons to celebrate. The second edition of the User Survey brought us more insights about usage of OpenStack around the world. We announced our first OpenStack Ambassadors, people who will help the OpenStack Community team get closer to many communities around the world. It was great to meet personally more women from the Outreach Program for Women of OpenStack. we’ve been doing this for over a year now, it’s a thing.

IMG_20131111_204451Growth also brings challenges and some of them were evident in some of the conversations had at the Summit, around it and after. A few signals I caught during and around the Design Summit sessions highlighted that we may need to start taking new steps to reinforce the culture of collaboration inside the project. The challenges highlighted go from lack of reviewers (not core reviewers, just developers who pay attention and help others), PTLs getting overloaded, the high traffic on the Development mailing list (which leads to loss of information), the increasing number of questions on Ask OpenStack with no interactions (no up/down votes, comments, etc) and little engagement in its Chinese version, the challenges inside the Internationalization Team with processes and tools. We’ve also heard of a very few Design sessions where it was too hard to have a productive discussion because of one or two uncollaborative people.

Since we’re getting so many new developers in the project we’re probably getting to the point where we can’t assume they are accustomed to contributing upstream first. The founders and first members of OpenStack all had a brilliant pedigree of open source contributions and collaboration. New members of the OpenStack Foundation may need some help to succeed. I enjoyed the session Getting Your Blueprint Accepted Quicker: the VPNaaS Use Case so much that I’m proposing the Upstream University training as an official program at the OpenStack Foundation to help new members. I’ll write more about this in the future.

The next six months will continue to be super exciting and full of things to do. If you have missed Hong Kong go watch the recordings of the sessions  keep watching this space for more news.

 

What Linus can do inside Linux community and you can’t do in yours

Linux kernel community has long been considered the greatest of all the open source communities. Linus Torvalds and his team has set the ground for open source development, defined processes and tools adopted and shared by other successful projects. The Linux kernel mailing list with its public review of patches and git, the tool to manage the incredible flow of code among thousands of people in tens of different branches laid the ground for many other open source projects. Even OpenStack doesn’t get close to those number (although OpenStack is only a three years old toddler and the kernel is old enough to vote and drink.) Linus has built an incredible community and an impressive culture around it. A culture where technology rules everything and also profanity and insults are common. And the results are clear: it worked for Linux kernel.

This doesn’t mean that any community can live and prosper like the Linux kernel with the same culture of harsh criticisms, middle fingers or what Linus calls management by perkele. In fact, I think nobody else can afford managing open source communities the way Linus does. Torvalds can get away saying things like “trying to come up with some ‘code of conduct’ that says that people should be ‘respectful’ and ‘polite’ is just so much crap and bullshit”. I certainly can’t and chances are you can’t either.

Now if you ask me if Torvalds should change his attitude my answer is: no, he is what he is and he’s made what he’s made because (or despite, who cares: results matter) of what he is. Should his lieutenants be assholes too? Of course not, and that’s why the kernel is still one of the most successful open source projects out there.

via Linus Torvalds defends his right to shame Linux kernel developers | Ars Technica.