GitHub blames record traffic, not a bad deploy, for its 7-hour 47-minute outage

GitHub's post-mortem pins the August 17 outage — 7 hours 47 minutes across Actions, pull requests, issues, Copilot and APIs — on a Central US component that failed to scale, not on any code or configuration change. Traffic is the story: 2.9 billion commits a month, up from 1.4 billion in April, plus 24 million new repos, with VS Code's retry behavior amplifying the failure. CTO Vladimir Fedorov's remediation list — isolate critical systems, tighten retry limits, alert on traffic spikes — describes the retry-storm failure mode any tool that auto-retries against a degraded API can reproduce.

Source: theregister.com ↗