Expo Outage History

EAS Workflows were stuck in a pending state with no logs for 2.2 hours due to a database table lock that prevented workflows from being processed. The lock was cleared and workflow processing resumed, but affected workflows were marked as errored and need to be re-run. The team is investigating the root cause to prevent recurrence.

Expo experienced login issues affecting some users accessing the EAS (Expo Application Services) through the website for 5.9 hours. Users were unable to authenticate properly, though clearing cookies for expo.dev provided a temporary workaround for some affected users. The incident was resolved after implementing a fix and monitoring the results.

A backwards-incompatible change was accidentally deployed to Expo's production environment, causing fingerprint jobs to fail due to a new requirement for `params` to be provided on the job. The incident affected fingerprint job processing for 22 minutes until a fix was deployed and the issue was resolved.

Some EAS builds failed to start due to an identified system problem, preventing users from successfully initiating their build processes. The incident affected the EAS Build component and lasted approximately 1.1 hours. The issue was resolved and affected builds were automatically waived to avoid charging users fees.

The EAS Build dashboard experienced an issue where users were unable to view their build pages. This affected the EAS Build component of the Expo service for 50 minutes. The incident was resolved after investigation and monitoring.

Expo's EAS Workflows service experienced increased queue times for 1.7 hours due to reaching worker capacity limits. Customers encountered longer wait times when submitting Workflow jobs during this period. The incident was resolved by addressing the capacity constraints.

Expo experienced elevated queue times for free tier users on Workflows and EAS Submit services for 1.8 hours. The incident affected EAS Submit functionality, causing longer than normal wait times for job processing. The team implemented measures to reduce queue times, and the maximum queue time steadily decreased until resolution.

Expo experienced an 18-minute service incident where new login sessions created after 11:22 AM PT were unable to access Launch (launch.dev). The issue prevented users from logging in to use the Launch service. The problem was identified and resolved, with all login sessions restored to normal functionality.

Expo experienced an increased rate of workflow runs failing to start due to an "Invalid workflow definition" error where the system expected a record but received undefined values. The issue affected users' ability to execute their workflows for 7 hours. The incident was resolved, with a temporary workaround involving minor YAML file modifications to trigger config re-instantiation from GitHub.

Expo experienced degraded push notification delivery for both Android and iOS platforms, affecting the Push Notifications Broker component. The service delays lasted approximately 10 minutes before being resolved. All queued notifications that were delayed during the incident were successfully delivered once the issue was fixed.