GitHub Availability Report: April 2026

GitHub recorded 10 incidents that caused service degradation over the course of April. In particular, the major incidents on April 23 and April 27 were disclosed in separate blog posts, and GitHub is also reflecting more details on its status page to increase transparency.

• On April 1, Code Search experienced a full 2 hour 20 minute outage after excessive automated changes during a messaging system upgrade coincided with an internal routing state reset, and during that window 100% of search requests failed. Stale results continued to appear afterward, but the underlying Git repository data was not corrupted, and full recovery was completed at 23:45 UTC.
• On the same day, Audit Log lost connectivity to its backend storage due to a credential rotation failure, making lookups unavailable via both the API and web UI for 28 minutes. No events were lost, and GitHub Enterprise Cloud with data residency customers were not affected.
• On April 9, a rate limiting bug in Copilot coding agent was applied globally across all users, significantly delaying new session creation, with queue wait times increasing to as much as 54 minutes. The average error rate was 83.9%, peaking at 97.5%, resulting in a cumulative 22,700 delayed or failed workflow creations.
• On April 13, GitHub Pages produced approximately 17.5 million HTTP 500 errors after an automated DNS management tool incorrectly deleted records for backend hosts.
• On April 16, Codespaces connections from VS Code failed due to upstream download failures, causing about 40% of startup tasks to fail. SSH connections were not affected.
• On April 20, a serialization error affected code scanning default setup, code quality, and project boards, blocking analysis of new PRs and issue updates, with recovery for project boards continuing into the next day.
• On April 22, Copilot Chat and Copilot Cloud Agent were disrupted due to a database connectivity issue, and the Copilot Memory preview was also unavailable.
• On April 23, unstable DNS infrastructure in one data center affected 5–7% of total traffic across Copilot, Webhooks, Git Operations, Actions, Migrations, and Deployments. The cause was a recently introduced traffic balancing mechanism that triggered DNS resolver failures under certain load patterns, and no data was lost.
• On April 27, a load balancer in front of search became saturated with anonymous scraping traffic, causing intermittent search failures across Issues, Pull Requests, Projects, Repositories, Actions, Package Registry, and Dependabot Alerts. This traffic accounted for 30% of daily search traffic and originated from more than 600,000 unique IPs.

Prompted by these incidents, GitHub stated it will strengthen more gradual rollouts, stronger health checks, deployment safeguards, DNS recovery automation, rate-limit scope separation, scraping blocking, and traffic isolation.