DEVOPS WEEKLY ISSUE #239 - 2nd August 2015

3 minute read

A fairly operations heavy issue with this, with discussions of software in production, monitoring, configuration management, system packaging and a few tools to help building out infrastructure thrown in.

Sponsor

Devops Weekly is sponsored by Brightbox Cloud: the multi-zone cloud platform built for High Availability.

Try an SSD cloud server in 30 seconds with your £20 free credit…
http://brightbox.com/devopsweekly

Sponsored

[Webinar] How to Use DevOps & APM to Release Better Software Faster

Join us for a talk with Hasan Yasar, Technical Manager of Carnegie Mellon’s Software Engineering Institute, to learn the benefits of DevOps, automation and CD, where APM fits in DevOps, and why you need to monitor. Finally, find out all about auto-provisioning and deployment with Docker, Vagrant, Ansible, and more. Register now.
http://ow.ly/Q3waZ

News

An excellent balanced write up of some of the main issues with running Docker in production - covers the kernel requirements, filesystems, secrets, logging, image building and more. Required reading for anyone looking to run Docker at some level of scale.
http://sirupsen.com/production-docker/

An excellent set of questions for any security minded infrastructure developer to ask about different parts of a service-based architecture.
http://www.grahamlea.com/2015/07/microservices-security-questions/

A great high-level introduction to devops, focusing on the people and practices that we’re evolving and how that affects our selection of products.
http://readwrite.com/2015/07/29/devops-people-not-technology

I really like the idea of infrastructure code shipping with tests which the end-user can then use to verify that the code works as intended. These two posts demonstrate that, as well as getting that information into elasticsearch and KIbana.
http://annaken.blogspot.co.uk/2015/07/automated-serverspec-logstash-kibana-part1.html
http://annaken.blogspot.co.uk/2015/07/automated-serverspec-logstash-kibana-part2.html

Some observations on a subject that’s close to my heart - containers and configuration management. Mainly a discussion of the trade-offs involved in adopting new technology before the tooling matures.
http://txt.fliglio.com/2015/07/containers-and-cfg-mgmt/

Another set of posts in this excellent series on what and how to monitor various common open source components - this time digging into monitoring the Varnish caching server.
https://www.datadoghq.com/blog/how-to-monitor-varnish/
https://www.datadoghq.com/blog/how-to-collect-varnish-metrics/

Coming to Java and JVM monitoring from other software tends to present a bit of a learning curve. This post introduces some of the approaches and tools which should help get you started.
https://blog.dataloop.io/2015/07/25/monitoring-java-apps-with-nagios-graphite-and-statsd/

A post looking at two different layouts for a service based system system, one point-to-point and another based on an event stream. I’m glad someone asked the ESB question in the comments though.
http://www.thoughtworks.com/insights/blog/scaling-microservices-event-stream

A good post on different ways of addressing complexity in systems, and in particular in the context of Facebook why a focus is placed on making changes reversible.
https://www.facebook.com/notes/kent-beck/taming-complexity-with-reversibility/1000330413333156

A quick reminder that system packages have lots of properties which other approaches tend to miss.
http://www.krisbuytaert.be/blog/power-packaging-software-package-all-things

Events

Devopsdays Berlin is coming up on the 26th and 27th of October. Registration and the call for proposals are both open now, and the conference is also on the lookout for sponsors.
http://www.devopsdays.org/events/2015-berlin/

Velocity Europe is coming up on October 28th until the 30th in Amsterdam with the usual mix of solid operations and performance talks and tutorials. The discount code DOW20 should provide a hefty 20% discount too.
http://velocityconf.com/devops-web-performance-eu-2015

Tools

ActiveMQ Artemis is a multi-protocol, embeddable, very high performance, clustered, asynchronous messaging system. It provides high availability and wide range of clustering options out of the box.
https://activemq.apache.org/artemis/

Python has always had a hand in data analysis work, but mainly for smaller or medium sized data sets. Ibis aims to change that, with first class integration the Hadoop ecosystem, without compromises in functionality, usability, or performance.
https://github.com/cloudera/ibis
http://www.ibis-project.org/

Racker is an opinionated Ruby DSL for generating Packer JSON templates. The main advantage appears to be composition and reuse, handy if you’re maintaining a number of templates.
https://github.com/aspring/racker

Testinfra is a tool for writing unit tests for infrastructure, similar to Serverspec but implemented in Python. It supports a few standard privatives like files, services, packages and users, as well as featuring integration with Salt and Puppet.
https://github.com/philpep/testinfra
http://testinfra.readthedocs.org/en/latest/

Devops Weekly is sponsored by Brightbox Cloud: the multi-zone cloud platform built for High Availability.

Try an SSD cloud server in 30 seconds with your £20 free credit…
http://brightbox.com/devopsweekly

Updated: