DevOps
Another day, another GitHub wobble - but the service keeps growing
GitHub Actions, a service which builds, tests and deploys code, was down for more than three hours yesterday, accompanied by an alarming and incorrect error message that stated that "Your account is suspended."
The impact of an outage in the code shack's Actions service can be greater than one that affects access to code repositories, because while developers can continue working with code on their local machine, there is no quick way to avoid Actions embedded in a CI/CD (continuous integration/continuous deployment) workflow.
"As the person on call for Continuous Integration team in my company, using Github is very stressful. Our CI is currently basically blocked," said one impacted user.
REG AD
The outage was first reported by users at around 1030 UTC on May 26, though the official incident report began at 1057 UTC, describing "degraded performance for Actions and Pages."
REG AD
This description was later revised to state that "the majority of Actions runs is impacted,” with the cause attributed to authentication issues.
It is possible to configure GitHub Actions to use external or self-hosted runners – the VMs on which Actions execute – but customers with this kind of configuration still experienced an outage as the GitHub cloud service is the control plane for the runners wherever they are located.
Adding to the stress caused by the failure of a critical service was the alarming (though fortunately inaccurate) error message. "My action failed with 'Unexpected error fetching GitHub release for tag refs/heads/master: HttpError: Sorry. Your account was suspended'," reported one developer.
Actual account suspension by a cloud service provider can take days to resolve, and involves battling automated systems. In the ensuing discussion another dev remarked that "I recently got my GitHub account suspended for four months. When it was finally reinstated, their support just said it was a 'mistake'."
GitHub reliability has been poor this year, sometimes thanks to activity by AI coding and agents, and that of bots grabbing data for LLMs (large language models) to consume, and sometimes for other reasons: it is hard to blame authentication issues on the bots.
Following every major outage there is discussion of GitHub alternatives, with some organizations moving to self-hosted code repositories and/or CI/CD. GitHub is sticky though, partly because of its generous free offer, and partly because of the cost of migrating existing workflows.
As one observer on Hacker News noted, GitHub usage is growing, not declining. COO Kyle Daigle reported on X last month that "platform activity is surging. There were 1 billion commits in 2025. Now, it's 275 million per week, on pace for 14 billion this year if growth remains linear (spoiler: it won't.) GitHub Actions has grown from 500M minutes/week in 2023 to 1B minutes/week in 2025, and now 2.1B minutes so far this week."
The increased activity is likely related to AI coding generating huge amounts of code at speeds humans cannot match.
GitHub reported the issue as resolved at 1318 UTC, although it added that "a small number of Issues, PRs, Comments, and Discussions were marked as hidden. We are working on correcting the underlying records."®