Date: January 20, 2026
Duration: 30 minutes
Severity: Minor

None. Only newly deployed pods were affected and entered a crashloop state. Existing pods remained healthy throughout, so the service continued operating normally for all users.

A code change introduced a GraphQL resolver definition that referenced a field not present in the GraphQL schema. This mismatch caused new pods to fail during startup validation.

The problematic code change was reverted in both the application repository and the deployment system (ArgoCD), restoring normal pod deployment behavior.

What went well: Kubernetes restart alerts detected the issue quickly, allowing for rapid response before any customer impact occurred.

Prevention opportunities: This issue would have been caught by running local or staging environment builds before deployment, as the pods would have failed to start and the error would have been visible in deployment health checks.

Broader observation: Schema validation mismatches between resolvers and GraphQL definitions are a recurring risk. Automated pre-deployment checks that verify resolver-schema alignment could prevent similar incidents.