My Big Continuous Delivery Summary Post

I’ve accepted that I can’t help people roll out Selenium without actually helping them also roll out Continuous Delivery practices. As such, I’ve started to do talks and workshops on it. The first was a full-day workshop at STPCon and yesterday I recycled the same deck for a talk at a local conference.

**[Automation and the Deployment Pipeline](http://www.slideshare.net/agoucher/automation-and-the-deployment-pipeline "Automation and the Deployment Pipeline")**

View more [presentations](http://www.slideshare.net/) from [Adam Goucher](http://www.slideshare.net/agoucher)

Of course, the deck is pretty, but it doesn’t do much good without actually having me in front of it. So here are some notes around it. As I get more-and-more things to link, I will.

Title

• book – Continuous Delivery
• post – Camping Heuristic

An Ethical Distinction

• post – Continuous Deployment vs. Continuous Deliver
• post – Testing vs. Checking

The Lightsaber Heuristic

• post – The Lightsaber Heuristic

Lots of Moving Parts

• Acceptance criteria
• Configuration management
• Continuous Integration
• Data migration
• Deployment
• Functional test automation
• Immunization systems

Specifications

• Book – Bridging the Communications Gap
• Post – Specifications
• Post – Mindmaps
• Tool – Cucumber
• Tool – Robot Framework

Manual Inspection

• Idea – Clean trunk vs. Dirty trunk
• Post – Feature toggle(1)
• Post – Feature toggle(2)

Commit Phase

• Post – Policy enforcement with svn hooks
• Post – Empty commit message check
• Post – svn pre-commit for buddy
• Post – some random scripts you can have

Configuration Management – Most organization’s CM is stuff in the days of sail and schooners with lots of people involved doing lots of things to load/unload cargo. You want to move to the modern era with standardized, interchangeable containers. Once this is (fully) in place, you never should have to log into a machine to make a configuration change. And should you have to, then its a hole in your CM setup

• Tool – Puppet
• Tool – Chef
• Tool – CFEngine
• Tool – Managing the machines
• Tool – Selenium and Puppet

Continuous Integration – /Everything/ is run from here; job runs, deployments, etc. and is the Information Radiator for the whole organization.

• Tool – Jenkins
• Tool – Bamboo
• Post – Basic Jenkins Job Chains
• Post – Job Chain Starting Points
• Post – Parallel Job Chains in Jenkins
• Post – The CI Server as a Se-Grid replacement

Data Migration – Again, no more running SQL by hand! All scripted and version controlled

• Tool – DB Deploy
• Tool – ActiveRecord
• Book – Refactoring Databases

Deploy

• Tool – Fabric
• Tool – Capistrano
• Post – Controlling Apache nodes –

Functional Automation

• Tool – Selenium
• Post – Locators (see the comment I left)
• Post – Synchronization
• Post – Elements and Actions

Immunization Systems – Aot a lot of information on this, but essentially, have your monitor system check business metrics for a new build once pushed to an environment. If it drops below the mean for a specific amount of time, then roll it back

Organizational Changes – Continuous Delivery is not just a technical change, but an organizational one as well

• Post – G-Forces