GitHub experienced degraded performance across Actions and Pages on Thursday, with the incident now resolved. At its peak, approximately 96% of GitHub Actions runs on GitHub-hosted runners failed to start, while Pages builds also experienced failures. GitHub Pages remained accessible throughout the incident.
GitHub provides cloud-based development tools and hosting services relied upon by millions of developers and organizations worldwide for code storage, collaboration, and automation.
The incident began with reports of degraded Actions performance, initially affecting approximately 5% of runs with delays exceeding 5 minutes. The scope expanded to 30% of runs experiencing similar delays, with some failing after exhausting retries and causing customers to exceed hosted compute concurrency limits. At one point, roughly 96% of Actions runs were failing to start. Copilot Cloud Agent and Copilot Code Review also failed to start for approximately 30 minutes during the incident. Recovery began gradually, with the system working through a significant backlog of queued jobs.
The incident has been fully resolved. GitHub stated that a detailed root cause analysis will be shared as soon as it is available.