5 minute read

I think this might be one of the most packed issues of Devops Weekly for a while. I’ve been on holiday for most of the week but still found myself falling over excellent content when I got back. Discussions of risk, useful tools I want to start using if I can find the time, upcoming events not to be missed, the opportunity to be a TV presenter for the amazing Opschool and a few interesting jobs for anyone on the lookout in London. Enjoy.


Devops Weekly is sponsored by Brightbox Cloud - launch cloud servers in multiple UK datacentres in seconds…


The excellent opsschool folk are starting to create video content, and are on the lookout for more topics and presenters. The next filming slot is going to be during Velocity, at the Santa Clara Hyatt, from Monday June 17th 2013 to Friday June 21st 2013. If you’re around and have an idea get in touch with them.

If you’ve ever thought of devops as a “throughput-optimizing, process-accelerating, agility-creating, workflow-throttling threat to your very bureaucratic existence” then this next post is for you. A useful (and suprisingly funny) reminder than some people will need convincing about the benefits of devops.

The nice folks at 10gen are are running a survey about how you collect, store and use your data. The information collected should be more useful the more people complete it, I’ll keep an eye out for the results being published too.

A good example of someone using the Dashing dashboard framework. Includes some code samples for getting information from Icinga, Twitter and Drupal but it’s also just a good example of what’s possible.

Knowing how to debug a problem with a running server is important, in order to fix the problem and also so software you build is easier to debug. This post covers a good list of things to check from a beginners point of view, looking at tools like history, free, dmesg, netstat, vmstat, pstree and lots more. A handy refresher or a good starting point.

Nice presentation all about the Etsy DevTools team. Goes into what they do and why they exist and runs through some of the tools they have developed to help Etsy ship code.

The topic of web operations moving from a craft to a science crops up regularly, and this post covers the science. Looking at the maths around mean-time-to-recovery and mean-time-between-failure and defining risk in IT systems. Ends asking people to share numbers like this to further research, although I’d be interested in traditional enterprises as well as the new technology titans.

Another post about managing risk, this one discussing “what is the risk that amazon will goo down”. Some interesting points, and if you’re interested in risk and human factors the references section contains some great reading material.

Good presentation on using the fire hose of logs and monitoring data for security and compliance purposes. The examples are from Etsy who use Splunk but lots of good ideas for most environments with payments, user accounts or a dynamic web application.

The Heroku Platform API is now available in beta. This is a nice HTTP API to controlling all the innards of Heroku. Everything from reading and writing configuration variables, creating new applications, adding or changing addons. Expect lots of interesting things to be built on this.

A video presentation about using Docker (the rather nice LXC container wrapper) to build your own Platform as a Service. It’s a very simple toy example but shows how little code is required to do something quite powerful with docker.

Scaling non-trivial applications across multiple data centres can be difficult, this post aims to document some of the main things to consider. If you’re looking to do this in the future this post might be a good starting point.

I’ve featured a few of these talks previously, but all of the excellent monitorama talks are now available in one place, both slides and videos. This was one of the most information packed two day events I’ve been to and I’m already looking forward to Monitarama.eu in Berlin later this year.

One persons experience getting started with Chef and Vagrant. Good notes and observations about what this particular team found useful and what they found difficult or frustrating. Worth reading if you’re new to the tools or if you’re writing tools for developers.


Shazam Entertainment is looking for an energetic technical manager with a viewpoint to lead its Service Engineering team. If you’re looking for a challenge to lead the team charged with building, managing and scaling the world’s number one audio recognition service we’re looking for you. We are on a phenomenal growth path adding over 2M users per week with an installed base of over 300M. We are focused heavily on BDD, automation, monitor (most) everything, and continuous deployment methodologies. We run a mixed technology stack using Java, Play, PHP, and C along with a dynamic mix of Amazon and hosted services. And we love our graphite graphs, the more the merrier! The position is based in London (Hammersmith), UK. If you’re not based here, but want to be, we’re willing to listen! If interested send a note to [email protected] and pop devops in the subject.

The UK Government Digital Service (where I work) is on the lookout for Technical Architects (closing 11th June) and Web Operations folk (closing 17th June). Both of these roles will involve you playing an important part in building and running a new breed of Government services, as well as helping to scale agile and devops practices across a huge organisation. Lots of interesting technical, people, process and security challenges await the successful candidates, along with lots of nice new colleagues in our central London offices. More details about the roles can be found at:


Monitorama EU, the conference for monitoring artisans, has a date confirmed, 19th and 20th of September in Berlin. Registration doesn’t open until tomorrow but the call for proposals is open now until July 1st.


Graph Explorer is an alternative dashboard for Graphite. It provides some nice tool to create interactive and real time graphs, group related metrics and compose graphs using regex, tags or other queries. Lots of screenshots and a video should give you an idea of what it’s capable of.

Omnibus packages help isolate applications from underlying dependencies. This project builds a package for Puppet, using the Omnibus installer from Opscode. The projects comes with a Vagranfile which makes building packages for different operating systems easy too.

If you’re using Vagrant you’re probably downloading package data over and over as you happily start and destroy boxes. Vagrant Cachier looks like a neat solution to speed this up, creating a cache of packages on your host machine which can be used by the guests. It supports multiple package managers too.