The August 17 outage
Points and comments are a snapshot, not live.
GitHub's 7-hour outage on August 17 was caused by a capacity failure as traffic reached a new peak.
The outage disrupted github.com, authentication, Actions, APIs, Copilot, and other services for 7 hours and 47 minutes. Traffic hit a new peak and a critical infrastructure component in the Central US data center failed to scale. Recovery required traffic rerouting, isolation of affected infrastructure, and staged service restoration. Copilot services took longer due to a client-side retry loop that increased traffic during recovery. Since April, monthly commits have grown from 1.4 billion to 2.9 billion. GitHub has added over 3 million CPU cores and 120 petabytes of storage, and accelerated migration to Azure, which now serves roughly 58% of platform load (up from 12% in May). Changes include applying consistent retry limits and budgets across service interactions, and reviewing lower-priority alerts for components that could fail during traffic spikes.
What commenters are saying
Many commenters are skeptical of GitHub's explanation, arguing that Azure is both the supposed solution and a source of reliability problems. One notes LinkedIn abandoned its Azure migration after four years. Others defend capacity failures as a known phenomenon where systems can hit hard thresholds-going from fine to broken with a small load increase. Several suggest alternatives: self-hosted GitLab or Forgejo for companies (costing about one to two weeks per year in maintenance), Codeberg for hobbyists, and Tangled or SourceHut. A commenter points out that GitHub's free tier is a massive cost sink Microsoft treats as a loss-leader, making disruption difficult despite dissatisfaction.