Open WebUI on OpenShift

Deploy Open WebUI for offline/air-gapped environments with vLLM and RHOAI inference endpoints.

Deploy

# Edit configmap.yaml for your environment, then apply it first
oc apply -f configmap.yaml

# Apply Namespace, PVC, Deployment, Service, and Route
oc apply -f open-webui.yaml

Or use Kustomize:

oc apply -k .

Verify

oc get pods -n open-webui
oc get route -n open-webui

Configure

1. Open the route URL
2. Create admin account (first user becomes admin)
3. Go to Admin Panel → Settings → Connections → OpenAI API
4. Add your vLLM endpoint URL and API token

Features

• Offline mode - No internet required, all external fetches disabled
• Server-side API - Connections made from pod (bypasses CORS)
• Self-signed certs - SSL verification disabled for OpenShift routes
• Persistent storage - User data stored on PVC

Notes

• Review configmap.yaml before deployment and adjust it for your environment.
• Change WEBUI_SECRET_KEY in open-webui.yaml before production use.
• This manifest is intended for a single-replica deployment backed by local SQLite/Chroma data on the PVC.
• For multi-replica or multi-worker deployments, Open WebUI recommends moving to PostgreSQL and Redis: Scaling Open WebUI.
• Environment variable behavior and supported settings are documented in Environment Variable Configuration.

Files

File	Purpose
open-webui.yaml	Namespace, PVC, Deployment, Service, and Route in a single manifest
configmap.yaml	Feature toggles, API settings, and offline config
kustomization.yaml	Kustomize configuration for configmap.yaml and open-webui.yaml

Cleanup

oc delete -k .