Victor Martinez

If you would like to know more about measuring your CI/CD for troubleshooting or improving your CI/CD services this talk can be interesting for you.

CI/CD at large scale, even at small scale, is hard and might affect your team’s mood, frustration with slow builds, or flakiness in your CI/CD. By using OpenTelemetry for each execution, test, deployment that happens in your CI/CD you can answer questions that are not possible at a glance. The main aim is to reduce the time that developers spend checking log errors, pointing them to the root cause of errors. Filter real verifications errors related to the CI/CD service and third-party systems. Debugging an error is as easy as going to the trace related where all the context information is accessible. In this talk, we want to show you what we have done and learned in the last year since we switched to using distributed traces for the CI/CD and implemented some integrations that help us to leverage our CI/CD.

When Everything as Code converges to automate/test your processes, in this talk we would like to discuss further our journey and our vision to handle our automation programmatically.

When we started to use Jenkins at Elastic, there was already a team providing the service, similar to the SaaS you could see nowadays, so teams were encouraged to handle their automation using the Configuration as Code paradigm with the Jenkins Job Builder tool, but at a certain moment of time this particular approach didn’t scale much, and that’s when we started to think about a more robust, testable and automated CI/CD ecosystem.

Then our journey started by applying the below principles and practices:

• Everything as Code
• Everything is Tested
• Documentation as Code
• Functional testing
• Continuous Improvement
• Continuous Deployment

All the above concepts are the ones that helped us in this journey and that’s the reason for this talk to share our experience and vision. Besides gathering from you your opinions and feedback.