In October, GitHub experienced two incidents that resulted in degraded performance across services.
Incident 1: GitHub Codespaces Authentication Outage — October 17 (10:59–13:48 UTC, lasting 2 hours 49 minutes)
Impact: 67% of users were affected during this period, preventing them from creating and starting Codespaces.
Root Cause: The regional authentication layer experienced throttling with a global third-party dependency due to increased load from onboarding a new Codespaces region.
Resolution: The team mitigated manually by reducing load on the external dependency.
Preventive Measures: GitHub is implementing regional-level caching and measures to enhance authentication service resilience.
Incident 2: GitHub Copilot Regional Outages — October 25, 09:13 UTC through October 26, 02:40 UTC (3 hours 27 minutes cumulatively)
Impact: Code completion functionality experienced multiple brief outages affecting a subset of users globally, with each region experiencing approximately 12-minute outages.
Root Cause: Nodes hosting the completion model were being upgraded by an automated process, causing the disruptions.
Resolution: GitHub disabled the automated upgrade behavior.
Preventive Measures: GitHub prioritized improvements to global load balancing during regional outages.