GitHub Services Face Widespread Degradation Due to Infrastructure Issues

GitHub reported widespread service degradations on April 27, 2026, impacting Search, Actions, and project management tools. The disruptions were linked to an infrastructure issue that caused intermittent failures and search timeouts for users globally. While a specific fix is being deployed for Copilot Codex agent runs, GitHub continues to investigate and mitigate the remaining service issues.

Key Points

• Multiple GitHub services including Actions, Issues, and Pull Requests experienced significant performance degradation starting April 27, 2026.
• An ongoing infrastructure issue led to widespread search failures, causing timed-out requests and workflow run errors.
• Copilot Cloud Agent jobs using the Codex agent were specifically impacted, prompting GitHub to advise users to switch agents temporarily.
• GitHub identified the cause of the Codex agent failures and began deploying a targeted solution by 17:26 UTC.
• As of the final update, GitHub continues to investigate and attempt mitigations for intermittent failures across the platform.

Sentiment

The community is overwhelmingly critical of GitHub's reliability and Microsoft's stewardship. The dominant sentiment is frustration bordering on resignation, with many users actively advocating for migration to self-hosted alternatives. While a few commenters offer more balanced perspectives about contributing factors like AI load, the overall tone treats GitHub's outages as an inevitable consequence of Microsoft's priorities and the Azure migration.

In Agreement

• GitHub's silent failures (showing empty PR lists instead of errors) are particularly dangerous and misleading for developers
• The frequency of outages has become unacceptable, with incidents occurring on a near-weekly basis
• The outage appears connected to broader Azure infrastructure issues, not just GitHub specifically
• GitHub's reliability has been in decline, with historical uptime data showing a clear downward trend
• The incident is impacting real business operations, with teams delaying deployments and losing productivity

Opposed

• Pre-Microsoft outages may have been under-tracked, so the apparent increase could partly reflect better incident reporting rather than worse reliability
• The recent spike in outages may be driven more by AI-induced load increases than by Microsoft's management specifically
• Self-hosted alternatives like Gitea and Forgejo lack some features and have their own limitations, such as permissions bugs and missing CI capabilities
• GitHub Actions remains the best CI option for open source projects, and syncing code to alternative forges doesn't solve the CI dependency problem