Member-only story
GitHub’s Quietest Big Outage in August Ran 10 Hours 42 Minutes
The 17 August outage hit about 29K organisations. GitHub’s own figures say the Actions incident ran longer
It is just after 02:00 UTC on 7 August 2026, and nothing on your team’s dashboards is red. Actions has been healthy since midnight, and the push you made that evening never triggered its workflow. Nothing errored. The event was simply gone.
Your team is made up; the missing events are not. At 02:03 GitHub posted that some push and pull request trigger events “were not processed and cannot be replayed”. The GitHub outage that got the coverage that month came eleven days later, on 17 August. This one, on 6 August, ran longer.
I took durations and follow-ups from GitHub’s August availability report, and timestamps from IncidentHub, an outage-monitoring service that archived GitHub’s status updates and quotes its root cause analysis (RCA).
The first failures hit at 15:05 UTC. Seventeen minutes later customers could see it on the status page. Core services recovered at 17:00, and that should have been the end. The incident was closed at 02:04, 10 hours 42 minutes after it opened.
