GitHub favicon

GitHub outage on 2026-10-29

github.com
October 2026 29 Minor incident

GitHub's Copilot cloud agent sessions broke for 2h 31m on 29 October, with failure rates of 1 to 2% and raised latency on creating and listing sessions. The fault sat inside GitHub's own agent sessions service, not in anything you run.

Started

10:19 UTC

Duration

Lasted 2h 31m

Source

IsDown

Next time

GitHub is degraded. Is your product working?

An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.

50 monitors for your own websites. No card.

What happened?

GitHub opened the incident at 10:19 UTC, reporting impacted performance for some GitHub services. By 10:22 it had narrowed this to the agent sessions service, with increased failures and latency on creating and listing sessions. That service backs cloud based Copilot agent sessions across web, mobile, VS Code, CLI and the GitHub Copilot app, so the disruption followed Copilot wherever it ran rather than sitting in one client. At 11:01 and again at 11:37 GitHub repeated the same failure rate, 1 to 2%, and said it was applying mitigations while monitoring for recovery. GitHub moved to monitoring at 12:33, saying the degradation had been mitigated, and resolved the incident at 12:49, a total of 2h 31m after it opened. GitHub said a detailed root cause analysis would follow once available, but none is attached to this record, so the underlying cause is not on the record yet. No components list was published alongside the incident. No user reports were filed in this window, which is consistent with a low, single digit percentage failure rate rather than a hard outage of the whole product. The incident sits inside GitHub's 90 day pattern of 60 incidents with a median duration of 1 hour 23 minutes, so this one ran longer than the typical GitHub incident but not by an order of magnitude.

Learning

UptimeRobot's own probes against GitHub's api-auth and raw-cdn components, run from the EU and NA regions, stayed at 100% uptime over 24 hours and 7 days through this window, and this incident was not registered against those checks. That is not a contradiction. The fault was inside the agent sessions service that powers Copilot's cloud features, a layer those two monitored components do not touch, and the failure rate GitHub described, 1 to 2%, is low enough that a steady external probe hitting a different endpoint every 60 seconds can easily sail through green while a narrower internal path degrades. This is the general shape worth recognising: a specific feature inside a large platform can fail while the platform's authentication and content delivery stay healthy, because they are different pieces of infrastructure answering different requests. If your product calls Copilot's agent sessions directly, your own uptime checks on your own endpoints would not have caught this either, since the breakage sat upstream in GitHub's infrastructure, not in anything you control. Your monitoring covers your half. UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.

4.7
stars out of 5
284+ reviews on

Start monitoring in 30 seconds.

There's nothing to install. No credit card required. 50 monitors for free.