Self-Hosting Litefuse

Self-host Litefuse when you need to run observability data inside your own infrastructure, connect to private model gateways, or keep runtime dependencies under your operational controls.

Litefuse can run as a compact Docker Compose deployment for evaluation and small installations, on Kubernetes through the official Helm chart, or as a production deployment that connects the Litefuse web and worker containers to managed Postgres, Redis, object storage, and Doris/SelectDB.

Deployment Options

Use Docker Compose when you want the fastest path to a working single-node Litefuse instance. This package starts the application together with Postgres, Redis, MinIO, and Doris.

Use Kubernetes (Helm) when you run Kubernetes. The official Helm chart deploys the application together with bundled Doris, PostgreSQL, Redis-compatible Valkey, and S3-compatible SeaweedFS, and each store can be switched to an external or managed service independently.

Use Production Deployment when your infrastructure team already provides managed Postgres, Redis, S3-compatible object storage, and Doris/SelectDB. In this mode, Compose only runs the Litefuse web and worker containers.

Architecture

Self-hosted Litefuse is built from a small set of services:

Postgres stores metadata, Redis powers queues and cache, object storage stores ingestion payloads, and Doris/SelectDB stores analytical event data. The web service handles the UI and API traffic, while the worker processes background jobs and queued ingestion work.

Production Notes

• Run all containers with TZ=UTC. Litefuse stores and processes timestamps in UTC; changing the container timezone can cause inconsistent time-based queries.
• Use HTTPS for NEXTAUTH_URL in production.
• Keep secrets in .env outside version control or inject them through a secret manager.
• Use managed Postgres, Redis, object storage, and Doris/SelectDB for production reliability.
• Keep private service endpoints in NO_PROXY when using an outbound proxy.
• Back up Postgres and object storage according to your infrastructure recovery objectives.

Reliability Checklist

• Configure health checks for web and worker containers.
• Keep the Litefuse web and worker images on the same TAG.
• Monitor Postgres, Redis, object storage, and Doris/SelectDB separately from the application containers.
• Test upgrades in a staging environment before updating production.
• Verify /api/public/health after every deployment.

Next Steps

Start with Docker Compose for local validation. For production, deploy on Kubernetes with the official Helm chart, or use the production deployment pattern once you have managed infrastructure ready.