GitHub’s automation platform Actions and its hosting service Pages are failing, according to an incident page on the service’s status site (archived version available here). Unfortunately, these are just the latest stability problems for a service that has had a lot of them lately. GitHub first reported the issue at 1522 UTC on Thursday, noting that Actions were suffering from “degraded performance” before noting about 20 minutes later that availability was also degraded.
A few minutes after the availability notice, GitHub explained that workflow runs were failing to start, or failing partway through execution, and that requests to the Actions REST API were returning errors. The Microsoft subsidiary said users were also seeing unexpected rate limits. An unnamed cause was mentioned as having been discovered in the same update, but that was some two hours ago, and the situation has yet to be resolved, with Pages being added to the list of affected systems shortly after GitHub said it was working on mitigating the problem with Actions.
“Engineers have applied a number of mitigations and are rolling out a further fix across all affected systems now,” GitHub said in its most recent update as of 1740 UTC. The company added that “Copilot code review, Copilot coding agent, hosted runners, and migrations using GitHub Enterprise Importer” were also apparently being affected by the issue, and that Webhook deliveries might be delayed as well.
Those problems are in addition to the earlier Actions workflow failures, timeouts, and degraded Pages availability. This is the latest in a long string of uptime issues at GitHub - one need only look at the status history page for the platform to see how bad things have become. A similar Actions outage occurred mere days ago on July 29, a month in which 26 incidents were logged.
In June, GitHub logged 23, an additional 23 in May, and 26 in April. We’re six days into August and the status page lists six total incidents for the month so far. Things have become so bad that many developers have considered moving their projects off the platform, with Ghostty developer Mitchell Hashimoto saying that the platform has become so unstable that it’s no longer a place for developers to host “serious work.”
The company apologized in April for the continued glut of outages. Based on the consistent number of failures over the past few months, it’s safe to say those apologies haven’t amounted to much. GitHub has placed part of the blame on a surge in the use of AI stressing its systems.
It promised structural improvements in June that GitHub SVP Jakub Oleksy said would “permanently remove failure modes.” Again, looking at the status history since June, those improvements don’t appear to have been deployed, or made a difference, quite yet. We reached out to GitHub to get more info on this latest issue, but didn’t hear back.
Originally published at theregister.com
