GitHub Outage History

GitHub experienced a major incident where Copilot Code Review (CCR) and Copilot Cloud Agents (CCA) failed to start for pull request comments, with users reporting that open pull requests were not being displayed. The incident lasted 2 hours and was resolved after engineers mitigated the degradation and confirmed that the Copilot services were starting again for pull requests.

GitHub experienced a major incident affecting Pull Requests for 3.7 hours, specifically impacting the creation of new pull request threads including line and file comments. While top-level comments, opening, merging, and other pull request operations remained functional throughout the incident, users were unable to create new threaded discussions on code reviews. The issue was resolved through mitigation efforts, with gradual recovery implemented after the root cause was identified.

GitHub experienced a major service disruption lasting 39 minutes that impacted performance across some GitHub services, specifically affecting users' ability to start and view Copilot Cloud Agent sessions. The incident was resolved after investigation and GitHub indicated a detailed root cause analysis would follow.

GitHub Actions experienced degraded availability for 2.4 hours, with ubuntu standard hosted runners causing high wait times and issues with queuing and running jobs. The service team applied a mitigation to resolve the queuing and execution problems, then monitored for stability before declaring the incident fully resolved.

GitHub experienced increased latency and failures affecting SSH-based Git operations for approximately 1.8 hours, while HTTP-based Git operations remained unaffected. The service degradation occurred between 14:00 and 16:10 UTC on May 5th, 2026, with users experiencing elevated response times and connection failures when using SSH for Git commands. GitHub identified the suspected root cause, implemented mitigation measures, and fully resolved the incident after a brief monitoring period.

GitHub Actions experienced degraded performance for 3.8 hours, with elevated queue times and job failures primarily affecting hosted runners in the East US region, impacting 8-10% of runs. The incident was caused by issues with GitHub's compute provider and affected both standard hosted runners and hosted runners with private networking. GitHub resolved the issue by applying mitigations for standard runners and working with their compute provider to restore capacity, with private networking users able to failover to other regions during the outage.

GitHub experienced a major incident with degraded performance affecting multiple services including Issues, Webhooks, Git Operations, Pull Requests, Actions, Packages, GitHub Pages, and Codespaces. The incident involved increased latency and timeouts across GitHub's platform, with some services experiencing degraded availability. The issue was fully resolved after 58 minutes, with services gradually returning to normal operation throughout the incident.

GitHub experienced a 62-hour incident where pull request search and indexing functionality was degraded, causing incomplete pull request results to appear in repositories and affecting pull request listings. The issue was caused by problems with GitHub's Elasticsearch cluster that resulted in missing search records, though no pull request data was actually lost. GitHub resolved the incident by reindexing the Elasticsearch indexes and repairing missing and stale search records, restoring full search and indexing functionality for pull requests.

GitHub Actions experienced degraded performance affecting hosted Ubuntu runners (ubuntu-latest and ubuntu-24.04), with up to 5% of jobs experiencing delays or failures due to capacity constraints. The issue was gradually mitigated over 3.2 hours, with impact decreasing from 5% to less than 1% of affected jobs before full resolution. Self-hosted runners and other hosted runner types remained unaffected throughout the incident.

GitHub experienced a 2.3-hour service disruption affecting Copilot Cloud Agent jobs using the Codex agent, which were failing after starting. The incident was resolved after GitHub identified the root cause and deployed a fix, with users advised to use alternative agents during the outage.