GitHub has published a post-incident report attributing this week's near eight-hour outage to a chain of infrastructure failures rather than a single cause. The company traced the problem to saturated load balancers in its Central US facility, a misconfigured autoscaling policy that failed to monitor a key concurrency limit, and a previously unknown retry bug in Visual Studio Code that massively amplified traffic to its Copilot Token Service, delaying recovery. The incident matters because it left developers unable to work normally across core GitHub services for hours, adding to a string of recent reliability problems and prompting renewed talk of rivals such as Cursor.

The outage began at 1328 UTC on 17 August and was not fully resolved until 2115 UTC, lasting seven hours and 47 minutes and causing elevated errors across Issues, Pull Requests, APIs, Actions and Copilot. An Istio sidecar hit its concurrency limit, but the autoscaling policy watched only the host service, not the sidecar, letting a cascading failure develop that was worsened by overly aggressive retry logic; engineers mitigated it by cutting gateway retries and blocking inbound Copilot Token Service requests. Most services recovered by 1636 UTC and Actions by 1803 UTC, but the Copilot Token Service, hit by a VS Code bug that amplified traffic roughly tenfold, was not fully restored until 2102 UTC, with recovery further hampered by scraping attacks on codeload endpoints; GitHub says it will fix the autoscaling policy, review retry limits and address the VS Code issue.