The useful question is what changes for users, developers or buyers, and whether the announcement stays industry context or becomes something people can actually use.

GitHub experienced six incidents in June 2026 that caused degraded performance across its services. Monolith traffic in Azure’s Central US region peaked at 45%, below the target of 50%, after a deliberate pause in scaling following a May 21 stability incident. The company restarted the ramp on June 17 with stricter stability checks, prioritizing controlled progress over speed. Git traffic in Azure grew from 30% to 43% but fell short of the June target due to deliberate decisions to avoid user latency, including routing only HTTP traffic and waiting for additional vPoP traffic.

Underlying infrastructure improvements included the extracted pull requests service, pullsd, now handling 100% of anonymous pull request reads, and reposd ramping to 50% of REST read traffic before being paused for Redis capacity work. The users service offloaded 500,000 queries per second from the primary database, and API rate limiting was 97% handled at the Gateway. Client-side database load shedding was tested on 5% of production traffic, and two-person confirmation was required for interactive production access and ChatOps changes.

On June 4, 2026, Copilot code review experienced elevated failures for 1 hour and 25 minutes, with 81.6% of requests failing at an average rate due to an incompatible dependency release. The issue was mitigated by removing the problematic dependency and redeploying the service, with recovery completed by 18:55 UTC. To prevent recurrence, GitHub pinned the dependency version, added compatibility checks, improved fast-failure behavior, and enhanced monitoring for review completion failures.

On June 8, 2026, unauthenticated users faced sustained HTTP 504 errors for 2 hours and 6 minutes when accessing pull requests and related pages, peaking at 34% of requests. The issue was caused by a surge in abusive anonymous traffic, degrading the dedicated pool of web servers. Mitigation involved targeted traffic blocks, and future improvements include better detection, emergency blocking deployment, and routing changes for mixed traffic endpoints.