Author: Jakub Oleksy (@jakuboleksy)
This September 2026 GitHub availability report details the incidents that affected our services over the month. We provide this information to keep our community informed and to share what we learn from each incident. As we navigate the evolving landscape of software development—now deeply integrated with AI-driven workflows and decentralized collaboration tools—our commitment to transparency and reliability remains paramount.
September 2026 Incidents
In September 2026, GitHub experienced several incidents that impacted availability. We classify incidents by severity and provide a summary of each below. We also outline the steps we are taking to prevent similar issues in the future.
Incident 1: Elevated errors on GitHub Actions (September 5, 2026)
On September 5, 2026, between 14:23 and 16:45 UTC, GitHub Actions experienced elevated error rates for workflow runs. Approximately 12% of workflow runs failed to start, and some in-progress runs were delayed. The root cause was a misconfiguration in a routine update to our job scheduling service, which caused a backlog in the queue. We rolled back the change and increased monitoring on the scheduler. To prevent recurrence, we are improving our canary deployment process for the scheduling service and adding automated rollback triggers.
Incident 2: Degraded performance on GitHub.com (September 12, 2026)
On September 12, 2026, from 09:10 to 11:30 UTC, GitHub.com experienced degraded performance, with page load times increasing by up to 300% for some users. The issue was traced to a database query that became inefficient after a schema change. We optimized the query and added additional indexing. We are also enhancing our query performance testing in staging to catch such issues before deployment.
Incident 3: Outage in GitHub Packages (September 20, 2026)
On September 20, 2026, from 18:05 to 19:50 UTC, GitHub Packages was unavailable for all users. The outage was caused by a failed storage node that did not failover correctly. We restored service by manually failing over to a redundant node. We are reviewing our failover procedures and implementing additional health checks to detect and respond to such failures more quickly.
Incident 4: Intermittent failures in GitHub API (September 28, 2026)
On September 28, 2026, from 07:00 to 09:15 UTC, the GitHub API experienced intermittent failures, affecting about 5% of requests. The issue was due to a memory leak in a service that handles API authentication. We deployed a fix and restarted the affected services. We are now conducting a thorough review of the service's memory management and adding more rigorous testing for memory leaks.
Summary
In September 2026, we had four incidents that impacted availability. We are dedicated to learning from these events and improving our systems. We appreciate your patience and trust in GitHub.
For more details on our incident history, please visit githubstatus.com.
via GitHub AI Blog