February 2023 Summaries

GitHub Actions offers a versatile platform for integrating end-to-end testing into workflows, leveraging a rich ecosystem of third-party actions and Docker containers. The article demonstrates how to execute browser tests using Cypress, an automation tool that simulates user interactions with web pages, and API tests with Postman's command-line tool, Newman, within GitHub Actions. By using Docker images, users can efficiently incorporate necessary dependencies and testing tools without manual setup. The process includes executing Cypress tests with the official GitHub Action, capturing results using JUnit reports, and handling artifacts like video recordings of test executions. For Postman, which lacks an official GitHub Action, the Newman Docker image is employed to run tests and generate JUnit reports, which are processed with the dorny/test-reporter action. The flexibility of GitHub Actions allows the execution of Docker images for unsupported testing platforms, making it an adaptable solution for continuous integration and deployment workflows.

GitHub Actions is a hosted service that facilitates the integration of unit tests into development workflows, enabling developers to verify code changes effectively. By utilizing third-party actions, developers can process test results and generate reports, which aids in debugging and improving software quality. The text explores two examples: testing Java applications using Maven and the test-reporter action, and testing DotNET Core applications using the DotNET Core CLI. Both workflows highlight the importance of deferring the response to failed tests to the test processing actions, ensuring that the test results are comprehensively reported even if there are failures. Although GitHub Actions is primarily an execution environment for verifying and building code, it offers useful reporting features that are expected to improve over time despite some limitations in tracking test results longitudinally. The post concludes by suggesting further exploration of end-to-end testing within GitHub Actions and emphasizes the synergy between GitHub and Octopus for enhanced deployments.

Lead times in software delivery have varied definitions depending on the context—whether it's Lean Software Development, DevOps, or another perspective—and understanding these differences is crucial for effective process optimization. The DevOps definition focuses on the time from code commit to production deployment, while Lean Software Development, influenced by lean manufacturing, measures lead time from requirement discovery to fulfillment. These different definitions highlight how lead times serve as a customer measurement, where the customer could be a developer, user, or car buyer, depending on the context. Furthermore, distinguishing between lead times and cycle times is essential; lead times reflect the customer's view of system performance, whereas cycle times help identify bottlenecks within the system. The Theory of Constraints, created by Eli Goldratt, emphasizes the importance of identifying and resolving system constraints to enhance performance. In software delivery, common constraints include large batch sizes, approval queues, and unnecessary hand-offs, which can be addressed by using both lead and cycle times to pinpoint and resolve bottlenecks. Ultimately, ensuring that lead time measurements align with the customer's perspective is crucial for making impactful improvements in software delivery processes.

GitHub Actions, an integrated CI as a Service within GitHub, offers a streamlined approach to continuous integration by eliminating the need for separate build servers and reducing infrastructure maintenance. It utilizes runners, which are GitHub-hosted virtual machines, to execute workflows triggered by events in a repository or external webhooks, allowing flexibility in automation. The service is particularly advantageous for users already on GitHub, as it provides a cohesive environment with fewer components to manage. Pricing is based on a pay-as-you-go model, offering free monthly minutes and storage, with additional charges based on usage. While GitHub Actions is not suitable for those using other code repositories, it does provide a vast array of community-made actions and workflows via the GitHub Marketplace, simplifying the setup process. The installation of actions involves copying code into a repository's .yml file, which enhances functionality without altering the user interface.

In the first post of the "Inside DevOps" series, Joanna Wyganowska interviews Jason Dunnivant, a Release Engineer at Olo, to explore his journey and insights into the DevOps field. Jason describes DevOps as an approach to automating tasks to allow software engineers to focus on development rather than manual operations. His career path, which began in the military and transitioned through tech support to network administration, eventually led him to his current role, where he leverages his psychology degree to understand user needs better. Jason discusses the challenges and rewards of DevOps, emphasizing the importance of staying open-minded and innovative. He shares best practices from his organization, such as using microservices for faster releases, implementing Configuration as Code to involve development teams in deployment processes, and maintaining compliance with the Sarbanes-Oxley Act. He advises newcomers to DevOps to learn by doing and recommends the book "Cloud Native DevOps with Kubernetes" for further reading. Jason also reveals a surprising personal fact: he is a nuisance alligator trapper with The Florida Fish and Wildlife Conservation Commission.

Version 6 of the Octopus Azure DevOps plugin introduces significant updates that enhance deployment automation and streamline processes by eliminating the need for the Octopus CLI and .NET Framework, replacing them with direct API interactions through TypeScript. This update supports NodeJS, reducing compatibility issues related to deprecated .NET versions. The plugin now requires Octopus Server 2022.3+ due to its reliance on the Executions API, which simplifies deployment tasks by offering explicit fields for supported settings. A new Go-based CLI has been introduced, and while the older C# CLI remains in use until mid-2023, the Go CLI supports more advanced features and integrates scripting concepts. The plugin separates tenanted and standard deployments, and introduces outputs for step chaining, which enhances task automation. The update also addresses naming inconsistencies and transitions package metadata to build information, with deprecated fields retained for continuity. The release aims to enhance user experience and efficiency, particularly for YAML pipelines, by providing a more robust and user-friendly deployment management system.

On January 25-26, 2023, Octopus Deploy experienced a significant service disruption affecting customers primarily in the US and EU, caused by issues during a planned DNS and routing provider migration involving Azure Front Door (AFD). The incident, lasting over 37 hours, was triggered by a misbehaving AFD profile, compounded by a network outage in Azure, internal routing problems, and complications from a new Web Application Firewall (WAF). Initial mitigation efforts included creating a CNAME record to redirect traffic through a www subdomain, but the incorrect IP address resolution persisted until Azure made internal changes to the AFD profile. Octopus Deploy has since engaged with Azure to prevent future occurrences, refined its incident response processes, and emphasized the importance of rollback-safe migration steps to minimize downtime and service disruption.

On January 25, 2023, Octopus Cloud customers experienced network connectivity issues due to an outage in Azure's internal network, impacting Octopus Cloud, Octopus.com/OctopusID, and the Cloud Control Center. The disruption, caused by a command on a Microsoft Wide Area Network router that led to re-computation issues across routers, resulted in long latencies and service timeouts from 07:05am to 12:43am UTC. Octopus engineers detected and declared the incident, updating the status page and pausing processes to mitigate impact while Azure worked on resolving the issue. The incident, resolved after approximately four and a half hours, highlighted the challenges of upstream cloud provider outages and prompted Octopus to enhance their monitoring, refine escalation processes with Azure, and improve communication strategies to better inform customers of service impacts.