GitHub detailed a challenging February 2026 in its latest availability report, outlining six separate incidents that led to degraded performance across its platform. The company acknowledged the significant impact these disruptions had on user workflows and confidence, pledging to bolster system resilience. According to the report, the issues affected critical services, including GitHub Actions and GitHub Codespaces.
Dependabot Degraded
Early February saw Dependabot struggling, failing to create approximately 10% of automated pull requests for over an hour on February 2nd. This was attributed to a cluster failover that incorrectly connected to a read-only database. The issue was resolved by pausing Dependabot queues and restarting failed jobs after traffic was rerouted.
GitHub Actions and Codespaces Outage
A major incident on February 2nd caused widespread unavailability of GitHub Actions hosted runners and GitHub Codespaces for nearly six hours. This outage, which began at 18:35 UTC, impacted all regions and runner types, leading to queued and timed-out Actions jobs. Services leveraging this compute infrastructure, including GitHub Copilot, CodeQL, and Dependabot, were also affected.
The root cause was identified as a loss of telemetry that triggered mistaken security policies on backend storage accounts within their compute provider. These policies blocked access to critical VM metadata, halting all VM operations. Mitigation involved rolling back the policy changes, with full recovery for Codespaces occurring on February 3rd.
Configuration Errors Cripple Services
February 9th witnessed two distinct periods of degraded availability affecting github.com, the API, GitHub Actions, Git operations, and GitHub Copilot. These incidents, totaling nearly three hours of disruption, were traced to a configuration change in a user settings caching mechanism. This change triggered a massive volume of cache rewrites, overwhelming shared infrastructure and causing cascading failures in the Git HTTPS proxy service.
GitHub is implementing optimizations to its caching mechanism to prevent write amplification and adding safeguards for faster rollbacks. They are also fortifying change validation processes and addressing underlying issues in the Git HTTPS proxy to enable automatic recovery.
Codespaces Provisioning Failures in EMEA
On February 12th, users in Europe, Asia, and Australia faced significant failures when attempting to create or resume GitHub Codespaces. Peaking at a 90% failure rate, the issue stemmed from an authorization claim change in a core networking dependency. Delayed detection was attributed to alerts not having adequate severity assigned.
Improved validation of backend service changes, enhanced monitoring during rollouts, and updated alerting thresholds are being implemented to prevent recurrence.
Repository Archive Download Errors
Later on February 12th, a brief 34-minute incident impacted users attempting to download repository archives containing Git Large File Storage (LFS) objects. This was caused by the deployment of an incorrect network configuration in the LFS service, though standard archive downloads remained unaffected.
GitHub's engineering and security teams are implementing system-wide improvements, drawing on lessons from incidents that echo challenges in broader Engineering practices at GitHub and the wider industry's approach to stability and reliability.
