Jenkins to GitHub Actions Migration Automation

4 posts · Jun 2021 – Dec 2024

What this trail covers

This thread tracks the evolution of Slack's CI infrastructure migration from Jenkins to GitHub Actions. Initial posts detail the challenges with Jenkins and the strategic decision to move to GitHub Actions. This post details the development and implementation of an AI-assisted conversion tool to automate the migration of Jenkins pipelines to GitHub Actions, significantly reducing manual effort and migration time. It covers the use of GitHub's importer tool, custom Python scripting, and LLMs for replacing rate-limited actions with internal mirrors and substituting specific actions like rtCamp/action-slack-notify with Slack's internal messaging action.

All Slack trails

Dec 2024Migration Automation: Easing the Jenkins → GHA shift with help from AIThis post details the development of an AI-assisted conversion tool to automate the migration of Jenkins pipelines to GitHub Actions. The tool leverages GitHub's Actions Importer, Python scripting, and LLMs. Specific AI-driven tasks include replacing rate-limited actions with internal mirrors and substituting third-party actions (e.g., rtCamp/action-slack-notify) with Slack's internal messaging action. The post outlines the planning, research into correction categories (rate-limited actions, action replacement, comments), and the implementation architecture of the corrections tool, emphasizing prompt engineering for AI corrections.
May 2024Balancing Old Tricks with New Feats: AI-Powered Conversion From Enzyme to React Testing Library at SlackThis post details the development and implementation of an AI-assisted conversion tool to automate the migration of Enzyme frontend unit and integration tests to React Testing Library (RTL). It describes the challenges of Enzyme's lack of React 18 support, the limitations of using Abstract Syntax Tree (AST) transformations alone (achieving only 45% conversion rate), and the inconsistencies of using Large Language Models (LLMs) independently (40-60% success rate). The post then introduces a hybrid approach combining AST and LLM transformations, augmented with DOM tree collection, which achieved an 80% conversion success rate. It also covers the open-sourcing of this conversion tool.
Jul 2022AutoTransform: Efficient Codebase ModificationThis post introduces AutoTransform, an open-source framework for large-scale codebase modification, designed to reduce the cost of maintaining large codebases. It details the framework's schema-based and component-based architecture, automatic scheduling, and change management capabilities. The post also outlines plans for future development, including open-sourcing more modifications and components, and adding functionality for running against human-authored changes in CI.
Jun 2021How a Jenkins Job Broke our Jenkins UIThis post details a specific debugging incident where a Jenkins UI became unresponsive due to a custom Groovy script that interacted with the Jenkins API to mark nodes offline. The issue was traced back to a recent CVE fix in the Script Security Plugin that tightened the Groovy sandbox, causing it to reject unsandboxed property gets from a custom `OfflineMessage` class. The fix involved replacing the custom `OfflineMessage` class with a simpler `OfflineCause.ByCLI` constructor that accepts a String message, thus avoiding the sandbox restrictions. It also provides takeaways on maintaining Jenkins environments, keeping API integrations lean, monitoring the Groovy sandbox and plugin updates, and having well-maintained runbooks.
Advertisement