GitHub experienced six significant incidents in February that degraded service performance. The company acknowledged the impact, stating: “We recognize the impact these outages have had on teams, workflows, and overall confidence in our platform.”
Incident 1: February 02, 17:41 UTC (1 hour 5 minutes)
Dependabot service failed to create approximately 10% of automated pull requests from January 31 through February 2. A cluster failover connected to a read-only database caused the degradation. GitHub paused Dependabot queues and restarted failed jobs. The company added monitoring and alerts to improve future detection.
Incident 2: February 02, 19:03 UTC (5 hours 53 minutes)
GitHub Actions hosted runners and Codespaces became unavailable between 18:35-23:10 UTC (with regional variations extending to February 3). Affected services included standard runners, larger runners, Codespaces, Copilot coding agent, code review, CodeQL, Dependabot, Enterprise Importer, and Pages.
Root cause: telemetry loss triggered automatic security policies blocking VM metadata access on compute infrastructure. GitHub mitigated by rolling back policy changes starting 22:15 UTC.
Incidents 3 & 4: February 09 (Two separate events)
Two related degradation periods affected github.com, the API, Actions, Git operations, and Copilot:
- First: 16:12-17:39 UTC (1 hour 27 minutes)
- Second: 18:53-20:09 UTC (1 hour 16 minutes)
A user settings caching mechanism configuration change caused simultaneous cache rewrites, overwhelming shared infrastructure. The first incident involved asynchronous rewrites; the second involved synchronous writes causing replication delays. Both led to connection exhaustion in Git HTTPS proxy services.
Mitigations included optimizing the caching mechanism, adding self-throttling, improving change validation, and fixing underlying connection exhaustion issues.
Incident 5: February 12, 07:53 UTC (2 hours 3 minutes)
Codespace creation and resume operations failed across Europe, Asia, and Australia, peaking at 90% failure rates. An authorization claim change in core networking dependency caused provisioning failures. Alerts lacked appropriate severity, delaying detection. GitHub improved validation, monitoring, and automated failover mechanisms.
Incident 6: February 12, 10:38 UTC (34 minutes)
Repository archive downloads including Git LFS objects failed between 09:16-11:01 UTC. Error rates averaged 0.0042%, peaking at 0.0339%. Incorrect network configuration in LFS Service caused health check failures. GitHub manually applied corrected settings and added corruption checks with auto-rollback detection.
GitHub directs users to its status page for real-time updates and references a companion blog post detailing root causes and resilience improvements.