GitHub has had a rough week. Its own status page lists 12 incidents between 2 and 9 November 2026, touching Git operations, pull requests, Issues, Actions, Pages, webhooks, the API and Copilot. Two more started on 9 November alone, and a "GitHub Down Again?" thread reached 121 points and 131 comments on Hacker News.
The part that broke builds: release downloads. From the morning of 8 November (users report from about 08:45–09:45 UTC), requests to github.com/<owner>/<repo>/releases/download/... returned 504 Gateway Time-out. Failing files reported in GitHub's community forum included the GitHub CLI, git-lfs, cosign, a Temurin JDK and a Grafana Terraform provider. Users there said this hit GitHub-hosted Actions runners, AWS CodeBuild and several regions, and that Gradle, actions/setup-java, cosign-installer and OpenTofu installs failed. In the setup-bun repository, a user reported that Bun downloads failed from about 10:00 UTC and worked again around 13:45 UTC; that also broke claude-code-action, which installs Bun. GitHub logged an incident for "timeouts for some unauthenticated users" from 10:36 to 12:32 UTC, while some users said the status page had shown everything green earlier.
9 November: GitHub opened "Increased timeouts due to surge in anonymous users" at 05:59 UTC. It said a spike in anonymous traffic caused 504 errors for people who were not logged in, CI jobs and release downloads; logged-in users and authenticated requests were not affected, it said. The errors came back after a first fix and the incident was resolved at 09:09 UTC. At 10:00 UTC a new incident started with delayed Actions run starts and webhook deliveries; at 10:47 UTC GitHub said a mitigation was applied and it was monitoring recovery.
Earlier in the week: on 3 November, Git over HTTPS was slow from 15:00 to 22:37 UTC (median response times up from under 100 ms to as much as 5 seconds for about half the traffic), which GitHub blamed on excess connections saturating its HTTPS hosts. On 4 November, a configuration change to an internal event-routing service paused event delivery twice, delaying Actions runs by up to 25 minutes and cutting Pages deployments to about 4% of normal. GitHub has not yet published a root cause for the 8–9 November problems.
