In June, GitHub experienced three incidents affecting service availability.
June 5 Incident (17:47 UTC, lasting 1 hour 33 minutes)
Between 17:47 and 19:20 UTC, GitHub Actions was primarily affected: 47.2% of runs had delayed starts of 14 minutes on average, and 21.0% of runs failed. Copilot Coding Agent had 60% of sessions cancelled, and GitHub Pages branch-based deployments failed.
Root Cause: A spike in load between internal Actions services exposed a misconfiguration that caused throttling of requests in the critical path of run starts.
Resolution: GitHub corrected the service configuration to prevent throttling and updated its deployment process to ensure the correct configuration is preserved.
June 12 Incident (17:55 UTC, lasting 3 hours 12 minutes)
Between 17:55 and 21:07 UTC, GitHub Copilot (Gemini and Claude models), VS Code, JetBrains IDEs, and Copilot Chat experienced significantly elevated error rates for chat completions, slow response times, timeouts, and chat functionality interruptions.
Root Cause: An outage affecting one of GitHub’s model providers.
Resolution: GitHub temporarily disabled the affected provider endpoints.
Preventive Measures: GitHub committed to updating its incident response playbooks for infrastructure provider outages and improving monitoring and alerting systems.
June 17 Incident (19:32 UTC, lasting 31 minutes)
Between 19:32 and 20:03 UTC, GitHub’s UI (3-4% authenticated error rates), API (40% authenticated errors, ~100% unauthenticated errors), Actions (2.5% delayed, 3% failed), and Large File Storage (1% errors) were all affected.
Root Cause: An internal routing policy deployment to a subset of network devices caused reachability issues for certain network address blocks within GitHub’s datacenters.
Resolution: The deployment was rolled back at 19:54 UTC, with full restoration by 20:03 UTC.
Preventive Measures: GitHub is expanding its validation process for routing policy changes.