GitHub Availability Report: April 2023

In April, GitHub encountered several service disruptions, including incidents affecting GitHub Copilot, Packages, and Codespaces, with investigations into their causes still ongoing. A notable event on March 27 involved a change in a frequently-used database query that led to lock contention and resource exhaustion, requiring manual intervention to recover services. On March 29, a degraded database cluster impacted GitHub Actions, attributed to a new load source and underprovisioning of proxy instances, which was mitigated by throttling job processing and adding capacity. Another incident on March 31 resulted from a misconfigured notification expiration date, causing 500 errors for some users, which was resolved by deploying fixes and auditing notification configurations. An outage on April 18 was linked to a planned database infrastructure change, which briefly disrupted access to issues and pull requests but self-healed after 11 minutes. GitHub has taken steps to enhance monitoring, alerting, and change management processes to prevent future occurrences.