GitHub availability report: July 2026
GitHub reported eight service incidents in July 2026, including a prolonged GitHub Actions outage on August 6, prompting accelerated migration to Azure and architectural improvements to enhance resilience and reduce shared infrastructure dependencies.
GitHub experienced eight incidents in July 2026 that degraded service performance, with a particularly impactful GitHub Actions outage on August 6 lasting several hours. The company acknowledged falling short of its availability commitments and is conducting a root cause analysis, with findings to be published in September. Immediate repairs and an accelerated architectural roadmap for GitHub Actions are underway, including a planned migration to Azure to improve capacity and resilience. The incident highlighted the need for better isolation and failover mechanisms to prevent cascading failures.
GitHub reported progress in reducing reliance on shared infrastructure, with over half of monolith read traffic now running in Azure Central US as of July 28. Git traffic in Azure reached 47%, up from 43% in June, and 29% of repositories now have a second replica in Central US. Stability validation processes introduced after a May incident are now standard for major traffic expansions, helping increase capacity without increasing risk. Authentication data has begun migrating from shared databases to dedicated infrastructure, reducing pressure on critical workflows.
The company implemented changes to improve reliability, including halving query time on the artifacts table and reducing Git authorization service load by 18.2% through caching. Search workloads now serve entirely from Central US with additional headroom, and GitHub is measuring the health of key customer workflows like pull requests to detect degradation earlier. Automation and operational safeguards are replacing high-risk manual production activities to reduce dependency on human execution.
GitHub’s next phase targets 70% of read traffic and 30% of write traffic in Azure Central US by the end of 2026, aiming to move all production traffic out of datacenters. The eight July incidents are detailed in separate write-ups, outlining system performance, failures, and corrective actions. The company reiterates its priority order: availability, then capacity, then features, as it works to enhance resilience and reduce shared failure points.