Install With Kubernetes And Helm
Kubernetes with Helm is the recommended path for production deployments that need high availability and platform-managed scaling.
See Installation for how this compares to the other deployment types, and for the shared a10e.toml configuration reference.
Prerequisites
Section titled “Prerequisites”- A Kubernetes cluster (1.24+)
- Helm 3.8+
kubectlconfigured to access your cluster
The Helm Chart
Section titled “The Helm Chart”The repository’s Helm chart lives at helm/env-manager/ — it is deployed by path from this repository, not from a public chart repository. It ships three values files:
values.yaml— local Kind defaults (imagePullPolicy: Never, dashboard disabled by default)values-dev.yaml— dev EKS overrides (dashboard enabled, ALB ingress,app.aliengiraffe.ai/api.aliengiraffe.ai-style hosts)values-prod.yaml— production EKS overrides
The chart manages PostgreSQL itself as a StatefulSet with a PersistentVolumeClaim — you do not need to provision Postgres separately.
Install PostgreSQL And The Connector
Section titled “Install PostgreSQL And The Connector”From the repository root:
helm upgrade --install a10e-local ./helm/env-manager \ --namespace a10e-local --create-namespace \ -f helm/env-manager/values.yamlFor a production-style deploy, layer your environment’s values file on top and override secrets explicitly rather than committing them:
helm upgrade --install a10e-prod ./helm/env-manager \ --namespace a10e-prod --create-namespace \ -f helm/env-manager/values-prod.yaml \ --set envManager.auth.jwtSecret="$(openssl rand -base64 32)" \ --set postgres.password="$(openssl rand -base64 24)"Key values worth reviewing before a production install (see the comments in helm/env-manager/values.yaml for the full set): image.repository/image.tag, envManager.auth.provider (local or point at WorkOS via the workos* fields), envManager.ai.*, postgres.storage/storageClassName, ingress.*, and dashboard.*.
Verify
Section titled “Verify”# Check deployment statuskubectl get pods -n a10e-prod
# View logskubectl logs -n a10e-prod -l app=a10eAll pods should reach Running. To check the health endpoint without exposing it publicly, port-forward to the API service and curl it — see Remote Access via SSH for the equivalent pattern on a bundle install, or use kubectl port-forward. See Operations: Health Checks & Backups for ongoing monitoring.
Upgrade
Section titled “Upgrade”helm upgrade a10e-local ./helm/env-manager -f helm/env-manager/values.yamlNext Steps
Section titled “Next Steps”Continue with Configure Data Sources, or see Installation for the full post-install checklist.