Delays with Actions Jobs for Larger Runners using VNet Injection in the East US region
This summary is created by Generative AI and may differ from the actual content.
Overview
On April 24‑25, 2026, GitHub Actions experienced delays and timeouts for Larger Hosted Runner jobs that used VNet injection in the East US region. The incident lasted from approximately 11:39 UTC on April 24 until 00:15 UTC on April 25, affecting only those runners without a failover region configured. Standard and self‑hosted runners were unaffected. The problem stemmed from backend failures in the compute provider’s provisioning, scaling, and update operations for virtual machines in the East US region. Mitigation was achieved by rolling back the affected Availability Zones, and a VNet failover capability was made available in public preview to allow evacuation of impacted runners.
Impact
The impact was limited to Larger Hosted Runner jobs with VNet injection in the East US region, which experienced increased latency, delays, and timeouts. Customers using these runners saw their workflows stall or fail, while other runner types (standard and self‑hosted) continued to operate normally. No broader service outage or revenue impact was reported.
Trigger
Backend failures in the compute provider’s provisioning, scaling, and update operations for virtual machines in the East US region caused the degradation of Larger Hosted Runner jobs with VNet injection.
Detection
The incident was first identified through customer reports of degraded performance for Larger Runners with VNet injection in East US, followed by internal investigation updates that confirmed the issue.
Resolution
GitHub mitigated the problem by rolling back the changes across all affected Availability Zones, restoring normal operation for the impacted runners. Additionally, VNet failover was released in public preview to enable customers to evacuate Larger Hosted Runners using VNet injection in similar scenarios.
Root Cause
The root cause was a failure in the compute provider’s backend systems responsible for provisioning, scaling, and updating virtual machines in the East US region, which disrupted the infrastructure supporting Larger Hosted Runner jobs with VNet injection.