GitHub October 2026 report: Azure capacity doubled but write outages and network disruptions persist

GitHub added substantial independently scalable Azure capacity, more than doubling compute across nine Central US Kubernetes clusters, while migrated-service traffic increased 82% since March.
Azure Central US now hosts 140 of 151 production services and 32 of 50 MySQL primaries, serving 63% of migrated-service reads, 66% of monolith reads, and 52% of Git reads.
Reliability improvements included a NAT-gateway mitigation that reduced usage by 85% and a database load reduction of 85% for a high-traffic Actions endpoint handling over 250 million requests per day.
October incidents included a 3-hour Actions queueing failure, a 19-minute write outage caused by outdated database routing records, and an Azure network disruption affecting all three Central US availability zones and at least 34,128 signed-in users.
The East US 3 buildout is approximately one week behind plan, with the first validated traffic ramp scheduled for the first week of December.

Incident Details

October 5: A partial networking failure caused GitHub Actions jobs to queue for up to 90 minutes, impacting GitHub-hosted runners and delaying GitHub Pages updates.
October 7: Two separate write outages totaling 19 minutes occurred when a new database orchestration machine published incorrect routing settings to retired servers.
October 29: An Azure network disruption affected all three Central US availability zones, impacting at least 34,128 signed-in users.
October 28: An invalid credential for an upstream model provider caused approximately 10% of requests to specific GPT-family models in Copilot to fail.

GitHub is implementing randomized delays for reconnects to prevent burst loads on overloaded servers.
Database routing settings now undergo stricter validation to catch incorrect configurations before they take effect.
The East US 3 regional recovery buildout continues with a focus on hardening migration safety and dependency isolation.