Incident with Actions, we are investigating reports of degraded availability
This summary is created by Generative AI and may differ from the actual content.
Overview
On May 6, 2026, from 06:45 UTC to 09:15 UTC, GitHub Actions Standard Ubuntu hosted runners experienced degraded availability, with 17.1 % of jobs failing. The degradation was caused by an unexpected data shape in the allocation‑configuration data for standard runners, introduced during remediation of a prior incident. Removing the offending data at 08:51 UTC unblocked allocations, allowing runner pools to scale and recover. Subsequent mitigation updated the filter logic to tolerate abnormal data shapes and added monitoring alerts for allocation blockages.
Impact
The incident reduced the reliability of GitHub Actions for users of Standard Ubuntu runners. Approximately 17 % of job requests failed, and wait times increased, leading to degraded service for affected customers during the three‑hour window. No broader outage of other GitHub services was reported.
Trigger
An unexpected data shape was introduced into the allocation‑configuration data for standard runners as part of post‑incident remediation work for the previous day’s incident. This malformed data blocked new runner allocations as load increased for the day.
Detection
Degraded availability was first reported by users and observed in monitoring telemetry showing high wait times for Actions jobs. The on‑call team began investigating the issue at 07:19 UTC after alerts indicated abnormal runner behavior.
Resolution
At 08:51 UTC the problematic allocation data was removed, allowing new runner allocations to proceed. A mitigation was applied to fix queuing and running of Actions jobs, the allocation filter logic was hardened against abnormal data shapes, and monitoring was enhanced to alert on allocation blockages. The system recovered fully by 09:15 UTC.
Root Cause
The root cause was the introduction of malformed allocation‑configuration data for Standard Ubuntu runners during remediation of a prior incident. This unexpected data shape prevented the allocation service from creating new runner instances, leading to blocked allocations and the observed degradation.