Choosing a Platform

Deploy chmonitor wherever your ClickHouse runs. Pick the platform that matches your infrastructure; each platform page is self-contained with install steps, all config categories, and migration notes.

Platforms

Platform	Best for	Auth options	Notes
Docker	Single-server self-host	none / clerk / proxy	Fastest path; pin a release tag
Kubernetes	Cluster-managed, multi-replica	none / clerk / proxy	Helm chart vendored in-repo; kustomize alternative
Cloudflare Workers	Edge / serverless	none / clerk / Cloudflare Access	D1 + Durable Objects for conversation store; Cron Trigger for health sweep
Vercel	Quick hosted preview (Next.js legacy app)	none / clerk / API key / reverse proxy	Use postgres or clickhouse for conversation store; no D1/DO
Node / standalone	Custom server, systemd, bare VM	none / clerk / proxy	bun run start or node server.js; put behind nginx

Configuration categories

Every platform shares the same configuration categories. The links below go to the reference page for each.

Category	What it controls	Reference
ClickHouse connection	Hosts, credentials, query timeout, pool	Configuration
Query / pool tuning	Execution time, cache TTL, pool size	Configuration
Feature permissions	Enable/disable features; public vs authenticated access	Features
Authentication	none / clerk / proxy (CF Access or trusted header)	Authentication
AI agent	LLM key, model, control tools, agent API token	AI Agent
Conversation store	Browser local / agentstate / D1 / postgres / clickhouse	AI Agent
Health alerting	Cron sweep, webhook URL, severity threshold	Configuration
Branding / analytics	Title, logo, GA / PostHog / Seline	Configuration

Not sure where to start?

• Running Docker already? → Docker
• On Kubernetes? → Kubernetes
• Want serverless edge? → Cloudflare Workers
• Behind Traefik / a reverse proxy? → Traefik
• Before going to production? → Production checklist