First community contribution

And there we have it, the first community contribution via git.centos.org is now live.

Andreas Thienemann’s offered to test and do some work of the work around the i686 CentOS-7 builds; and has gone through to propose mock configs that allow the i686 tree to be bootstrapped. I’ve just merged his pull request, and pushed it to the builders. The entire distro churn for i686 is now running, with details at seven.centos.org and you can see the build reports published in realtime at http://lists.centos.org/pipermail/centos-build-reports/2014-January/thread.html

This should allow everyone to quantify the scope of work we might be looking at, if as a community, we decide to maintain a i686 tree + installer for CentOS-7.

- KB

Talking to Scientific Linux

Hi,

Quite a few people have been getting in tough and asking if we are going to speak to the people at Scientific Linux to see how we might be able to collaborate with them. And I just wanted to say that yes, we are talking to them.

Its been quite a big change to the entire overall ecosystem and while we have had a few extra days to think about process proposals etc, they have’nt. So its going to take sometime to get expectations into proposals and into process; And we would like to have them come onbord in a way that works best for them, the organisation that backs them, and their user base.

- KB

We like ipv6

Here in the CentOS Project, we like ipv6.

In 2007 some of the private CentOS services started moving to ipv6; the entire buildservice has been ipv6 native since Sept 2008.

In 2009 we had mirrorlist.centos.org running over ipv6, which meant essentially the entire CentOS mirror network had ipv6 native support. This isnt something for dualstacked machines – even machines without an ipv4 address were completely supported.

In 2011 we had torrent.centos.org offering an ipv6 hosted tracker service, and a couple of seeds offering content natively on ipv6.

Now, in 2014 Jan, the new CentOS website has launched with complete ipv6 support. This includes the CentOS forums and all services hosted behind the www.centos.org site.

Most, if not all, this ipv6 work is driven by the efforts of Anssi Johansson.

And I also want to take this opportunity to thank two of our long term supporters who have provided us the resources to offer these ipv6 services. Coreix.net since 2007 and more recently UK2.net since 2011.

- KB

Git.centos.org is now open

As of about 30 minutes back, during the first ever CentOS OfficeHours session ( http://wiki.centos.org/OfficeHours ) we announced that git.centos.org is nw open to public access. In the coming hours, you will start to see content flow in as its injected from the older private setup into the new public setup. In order to hit the points that people care about most, as early as possible – I am going to get the content being CentOS-7′s builds available.

This includes the mock configs, the metadata around those, the distribution blacklists and whitelists as well as the distro-tree management scripts.

User ACL’s are complex and the actual contribution acceptable policy is going to hit the centos-devel list for comments once its been through a board review. But in order to not let that hold us all back from being able to jump in and start contributing, I’m going to get repo mirrors on github.com/CentOS, allowing us to start acceping pull requests right away. Lets start with that, and if its an easier route for most people, we can just retain the mirrors for inbound content going forward.

For now, enjoy!

- KB

As a community, for the community.

Ten years ago, a few people came together to solve a larger ecosystem issue. Red Hat Linux had split into the more commericial enterprise user oriented RHEL and the more fast paced, innovation across the board platform – Fedora. People familiar with the Red Hat way of running linux, who did not fit into those demographics had to make a choice. At the same time another group of people came up with and delivered on an idea : A community oriented distro, but built from Red Hat’s enterprise linux sources. CentOS was born, growing to satisfy that niche and evolving into a complete ecosystem of its own.

These ten years have seen CentOS grow to be the platform of choice in Web Hosting, Voice over IP, Cloud, Virtualisation, HPC and many many homebrew efforts. As contributors to this process, we are all proud to have helped facilitate the social web movement, been the center of the devops mindset, been a catalyst to many infrastructure deployment, management and monitoring technologies emerging in the last few years – and many more wins from all over the world. Today you can find CentOS Linux everywhere. In almost every computing demographic, there is a fair chance that if you run into a self supported, onsite tweaked linux install the its likely be CentOS Linux. The logo is widely recognised, and every cloud vendor of note offers CentOS Linux images.

The characteristic resilience of the platform, the energy in the community and widespread adoption have made CentOS Linux an exceptionally productive base to work on emerging technologies, allowing the developers and contributors to focus on the bits they cared about most while having a safety net around them that the base platform isnt going to rapidly evolve away. And people have used it, extensively; to develop on, to innovate on, to deliver better, more user friendly interfaces to a vast set of solutions.

However, this momentum around the platform has been in silos, away from the core – an effect that has meant the larger community is unable to come together in one place to further, faster, broaden the scope of development and stabilisation around that development. Added to that is the overhead of every silo having to bootstrap its own community. Those of you who have met me, at conferences or otherwise, would know that this is a problem I consider to be the biggest failure of the CentOS Project – the failure to grow organically beyond the platform. And it is very much a project level issue since it stems from the way we’ve had to work and the processs we have had to adopt. But these were the choices we have had to make because of where we are, the code we consume and how we consume it, then map that over to how we’ve tried to protect the userbase and the community that has given us their trust. And even when there have been opportunities to tap the community energy into CentOS itself, we’ve not always been able to that – back to the way we work, the way we do things, and why we do them.

Looking for an example of what I mean ? Consider for a moment that there are in excess of fifty, relatively well known, repositories of rpm packages that exist today targetting CentOS consumer space directly; none of whom consume or base their work out of .centos.org space.

The second biggest challenge we’ve faced has been the question of what-if the tap gets turned off. What if Red Hat becomes evil. Where does that leave the installed user base if all goes sour – and if it did, it will do so rapidly. Working and messaging around that threat isn’t easy, because it is very much a real and tangible threat. The small contributor base would never be able to either maintain or deliver on the established, expected user experience. Could we find the means to maintain it externally ? not easily and not without being disruptive to the existing user base – the people we care about most.

With the aim of addressing these issues, we are announcing today that the CentOS Project is joining forces with Red Hat. We can then make the next stretch of this exceptional journey alongside the giants whos shoulders we’ve stood on these last ten years. Our plan is to work together with the Red Hat OSAS team to remove the barriers we’ve had to build and live with in the past, create an open and publicly accountable meritocratratic governance group and to make the contribution process simple yet effective.

So how does this fit into the RHEL model and what does Red Hat have to say about this ? Personally, I still think the message we’ve promoted all along holds true even now: RHEL is a great commercial platform, over the last decade they have proven themselves repeatedly. The CentOS goal has always been to achieve a similar standing in the open communities. And its a goal that I will strive to ensure stays intact : To become the defacto community platform, by user choice.

Take a look at our official announcement at http://lists.centos.org/pipermail/centos-announce/2014-January/020100.html and come join us on irc, the mailing lists or the forums. I look forward to having a chat and answering any questions you might have about your CentOS story.

Here is a sampler of some of the things we hope to deliver in the coming months :
- A publicly accessible git repo that makes it easier for everyone to consume the code
- A SIG onboarding plan that allows anyone to propose and then curate a Special Interest Group
- An auditable process that allows a SIG to propose, build and then maintain their own CentOS Variant
- Better, bigger community outreach efforts
- Proposals to expand on the CentOS Extras/ and the CentOS Plus repositories
- A timeline with a tangible delivery plan, for the next 12 months
- Weekly board meetings, open to anyone who wants to come along and participate

And I’d like to welcome every one of you to come along with us, as we endeavor to expand the base platform into a bigger, more productive and more featuresome ecosystem. One that does not suffer the constraints of private privilege escalation and one that makes it possible for anyone to come along and help curate the bits they care about most. As a community, for the community.

Regards,

Karanbir Singh
kbsingh or z00dax on #centos-devel @ irc.freenode.net
tel: 0207.099.9389 / 0207.009.4455
Project Chair, The CentOS Project

Microsoft, Linux Integration Services and CentOS Linux

Interestingly a new version of the Linux Integration Services from Microsoft targetting RHEL and CentOS Linux. And bafflingly built for a Linux distro baseline otherwise obsoleted obout two years ago. Hopefully, a release that no one is running anymore.

http://blogs.technet.com/b/virtualization/archive/2014/01/02/linux-integration-services-3-5-announcement.aspx

Given that RHEL and CentOS Linux are considered stable – consistent, would it not make most sense for vendors like Microsoft targetting a larger consumerbase to just target EL-5 or EL-6 ? instead of a snapshot in time specific release iso …

In this case, who knows the best person to ping at Microsoft to pass on some clue ( or atleast workout why they are targetting something so old and obsoleted ).

- KB

An EPEL build in rhel7b1

Just as a PoC and to see how the inter-worker process locks work in the CentOS buildsystem, I churned the epel6 srpm set through rhel7b1′s public content. The entire run took just over a day, and I was surprised to see 4,440 binary rpms ( from 2,870 srpms ); There was no requeue / rpm-blame enabled here, and the packages were roughly built in the order they went in. 4,950 srpms were import from epel-6/srpms/30th Dec 2013.

Result is available here : http://mirrors.karan.org/epel7/ ; hope this further allows people to test the bits they care about in the upcoming release, and also something that the epel guys can find useful. For everyone who hasent seen the CentOS Buildsystem output, this is what it looks like. The entire contents of the mock resultdir are pushed, along with some metadata around the build, and every build has a ‘TAG’ which represents the date/time when it was ‘requested’.

Most people should/would want to track the real EPEL builds at : http://dl.fedoraproject.org/pub/epel/beta/7/

- KB