# Deployment

This page covers deployment concepts for Curiosity Workspace. The exact deployment model (desktop, VM, containers) may vary, but the production concerns are consistent.

# Deployment goals

Reliability: predictable uptime and recoverability
Security: controlled access, hardened configuration, audited operations
Scalability: handle data growth and query load
Reproducibility: promote changes from dev → staging → prod safely

# Environments

At minimum, maintain:

Dev: experimentation and rapid iteration
Staging: pre-production verification with prod-like config
Prod: restricted access, controlled changes

# What to version and promote

Treat these as deployable artifacts:

connector code
endpoint code
interface/front-end bundles
schema migrations and ingestion pipeline changes
search/index configuration
NLP pipeline configuration

# Deployment on OpenShift

Curiosity can be deployed on OpenShift using Docker images. To ensure data persistence, it is recommended to use a StatefulSet with attached persistent storage.

Prerequisites: A working OpenShift cluster and administrative access.
Docker Image: Use the official image from Docker Hub. For production, always use a versioned tag (e.g., curiosityai/curiosity:12345) instead of latest.
Storage: Configure a PersistentVolumeClaim (PVC) and attach it to your StatefulSet to maintain workspace state across pod restarts.
Configuration: Use environment variables to configure the workspace. Sensitive information like administrator passwords should be stored in OpenShift Secrets.

# Operational checklist (baseline)

Persistent storage with backups
TLS termination (directly or via a reverse proxy)
Secrets in a secret manager
Monitoring and alerting for:
- availability
- ingestion failures
- index rebuild failures
- latency spikes

# Next steps

Observe health and performance: Monitoring
Set permission boundaries early: Permissions