Start with what you have
Connect a Git repository, use a container image, or start from YAML and application templates.
Lightweight multi-server
control plane
Bring your cloud and home servers into one workspace. Deploy applications, trace their routes, and understand how they run.
Open source · Self-hosted · Built on Nomad
LUMA / CONTROL PLANE
Luma ControlDNS · TLS · Nomad
CLI / CI Console
Luma ControlDNS · TLS · NomadcnCloud server
globalCloud server
homePrivate node
Place by region. Connect by exposure.
Applications
Manage applications, service instances and releases.
Applications 4Healthy 4Degraded 0Failed 0
4 / 4
| Application | Status | Access address | Region | Node | Replicas | Actions |
|---|---|---|---|---|---|---|
| website1 service | Running | www.example.com | global | global-edge | 2/2 | |
| api1 service | Running | api.example.com | global | global-edge | 2/2 | |
| worker1 service | Running | Internal | cn | cn-edge | 1/1 | |
| storage1 service | Running | Internal | home | home-worker | 1/1 | |
| No matching applications | ||||||
Routes
End-to-end relationshipsFollow a service from its domain to its destination.
Configuration and instance relationships. Not a live connectivity probe.
Resource metrics
15 minA little context for every resource trend.
CPU42%Latest sample
Node usageIllustration · actual sample gaps remain visible
Delivery
Follow source, build and deployment results in one place.
websitemain ·
Deployed8d2a4f1- ✓Fetch source2s
- ✓Build image38s
- ✓Push image8s
- ✓Deploy application12s
Region globalReady replicas 2 / 2Duration 1 min
Observability
Locate affected services, inspect logs and track notifications.
Firing1
Enabled rules3
Notification channels1
Warning
worker · High memory usage
86% · Sustained for 5 minutes · Notification sent
Related service logs
12:15:03 worker [info] Processing batch 128
12:15:04 worker [warn] Memory usage above thresholdOverview
A view of your cluster, applications and releases.
| Resource | Status | Details |
|---|---|---|
| Applications | 4 running | 6 ready replicas |
| Nodes | 3 ready | cn · global · home |
| Latest delivery | Completed | website · main · 1 min |
Alert incidents
Follow active incidents, recovery and notification results.
| Object | Alert | Status |
|---|---|---|
| worker | Memory 86% · sustained for 5 min | Firing |
| api | Response latency returned to normal | Resolved |
Application observability
Connect service health, latency and error rates to applications.
| Application | Latency | Error rate | Status |
|---|---|---|---|
| website | 28 ms | 0.0% | Healthy |
| api | 64 ms | 0.1% | Healthy |
| worker | — | 0.0% | Memory alert |
Service logs
Inspect service output with instance and timestamp context.
worker / worker-01stdout · stderr
12:15:01 [info] Worker ready 12:15:02 [info] Connected to queue 12:15:03 [info] Processing batch 128 12:15:04 [warn] Memory usage above threshold 12:15:05 [info] Batch completed in 1.2s 12:15:06 [info] Waiting for next batch
Alert rules
Define sustained conditions for resource usage and service health.
| Rule | Condition | Duration | Status |
|---|---|---|---|
| Memory usage | > 80% | 5 min | Enabled |
| CPU usage | > 85% | 5 min | Enabled |
| Service unavailable | Ready replicas = 0 | 1 min | Enabled |
Notification channels
Review notification configuration and recent delivery results.
| Channel | Configuration | Latest notification | Status |
|---|---|---|---|
| Operations | Feishu · App Secret configured | worker memory alert | Delivered |
Infrastructure
Explore nodes, networking and storage across regions.
| Region | Node | Ingress | Storage |
|---|---|---|---|
| cn | cn-edge | HTTPS | Local volumes |
| global | global-edge | HTTPS | Local volumes |
| home | home-worker | Private relay | Local volumes |
Nodes
Node roles, resource usage and latest heartbeats.
| Node | Region | Role | CPU | Status |
|---|---|---|---|---|
| cn-edge | cn | Manager · Edge | 24% | Ready |
| global-edge | global | Worker · Edge | 42% | Ready |
| home-worker | home | Worker | 18% | Ready |
Join a node
Install node components, connect to Control and await the first heartbeat.
- 1Choose region
- 2Install components
- 3Await heartbeat
Target region globalNode role Worker
Regions
Regions define placement and their available nodes.
| Region | Nodes | Applications | Access |
|---|---|---|---|
| cn | 1 | worker | Internal |
| global | 1 | website · api | Public HTTPS |
| home | 1 | storage | Private relay |
System maintenance
Follow Control upgrades, node updates and route checks.
| Check | Result | Details |
|---|---|---|
| Public routes | Passed | 2 / 2 routes healthy |
| Control | Ready | Health check passed |
| Node versions | Aligned | 3 / 3 synchronized |
Storage
Persistent volumes retain their owning node across updates.
| Volume | Application | Node | Mount path |
|---|---|---|---|
| storage-data | storage | home-worker | /srv/luma/data/storage/data |
| worker-data | worker | cn-edge | /srv/luma/data/worker/data |
Images
Inspect application images and built revisions.
| Application | Image | Source |
|---|---|---|
| website | registry.example.com/website:8d2a4f1 | Git build |
| api | registry.example.com/api:3b7c912 | Git build |
| worker | registry.example.com/worker:stable | Container image |
Initial setup
Start with configuration and check deployment dependencies.
| Setting | Example value | Status |
|---|---|---|
| Cluster domain | example.com | Configured |
| DNS | Cloudflare | Connected |
| Certificates | ACME | Configured |
| Private network | Tailscale | Connected |
Settings
Manage deployment credentials and platform configuration.
| Configuration | Count | Purpose |
|---|---|---|
| Secrets | 3 | Environment references |
| Registry credentials | 1 | Private image pulls |
| Git accounts | 1 | Repositories and builds |
Secrets
Inspect names and usage; sensitive values are never displayed.
| Name | Reference | Applications |
|---|---|---|
| DATABASE_PASSWORD | ${DATABASE_PASSWORD} | api |
| SESSION_SECRET | ${SESSION_SECRET} | website |
| WORKER_TOKEN | ${WORKER_TOKEN} | worker |
Registry credentials
Store pull credentials for private container registries.
| Registry | Username | Credential | Purpose |
|---|---|---|---|
| registry.example.com | deploy-bot | Configured | Application image pulls |
Git accounts
Connect source providers and select repositories for delivery.
| Provider | Account | Repository | Connection |
|---|---|---|---|
| GitHub | example-team | website · api | Connected |
| Gitea | example-team | worker | Connected |
Storage configuration
Review local storage conventions and data ownership policy.
| Setting | Policy |
|---|---|
| Default backend | Local storage |
| Data directory | /srv/luma/data/<application>/<volume> |
| Placement | Pinned to the data owner |
| Application removal | Retain data by default |
Maintenance settings
Use System maintenance under Infrastructure for maintenance operations.
Control and node maintenance
Check the route baseline, then follow image preparation and updates.
Designed for everyday operations, not just the first deployment.Console guide
From source to service.
Keep the details in view.
Understand the path
Keep DNS, HTTPS ingress and private relays connected to the application they serve.
Stay close to runtime
Inspect instances, resource history and logs. Manage alert rules and notification channels in the same console.
Bring your deployment workflow to your AI assistant.
Give your coding assistant Luma configuration rules, deployment workflows and troubleshooting guidance.
Deployment and configuration
luma-deployment-yamlGenerate and validate YAML, review Compose configuration, and follow the application’s recorded build and deployment workflow.
Review this Compose project, generate its Luma sidecar, and validate it.View skill →
Application observability
luma-observeDeploy and operate the independent observability stack, checking application metrics, allocation alerts and notifications.
Set up luma-observe for this cluster and verify alerts and notifications.View skill →
A control plane.
On your infrastructure.
Luma coordinates established tools. Control holds configuration and orchestrates deployments; your nodes run the workloads.
Explore the architectureLuma Control
Configuration, credentials and deployment coordination
CONTROLNomad
Scheduling and workload lifecycle
COMPUTETraefik + Cloudflare
HTTP/HTTPS routing and DNS records
INGRESSTailscale
Private node connectivity and home relays
AS NEEDEDYour first deployment,
one step at a time.
Prepare a Linux manager, a domain and a Cloudflare DNS token. Add private networking when your topology needs it.
Requirements & full setup guide Ubuntu 22.04+ Docker / Nomad Cloudflare DNS
Install the CLI
Run on the manager first. The installer creates an isolated Python environment.
curl -fsSL https://raw.githubusercontent.com/LiuTianjie/luma/main/scripts/install-luma.sh | sh
~/.local/bin/luma preflightInitialize your manager
Configure DNS credentials first, then bootstrap the control plane.
luma bootstrap manager --domain luma.example.comLog in and deploy
Use a management token from a trusted client. Choose a manifest or use the console.
luma login https://luma.example.com --token <management-token>
luma deploy service.yamlThe next step, explained.
The console
Deploy, inspect and operate from one workspace.
Read guideInstall & bootstrap
Prepare the manager and join your first nodes.
Read guideDeployment configuration
Regions, ingress, resources and service manifests.
Read guideCompose & storage
Run multi-service apps with persistent data.
Read guideMetrics, logs & alerts
Understand what is running and investigate issues.
Read guideCLI reference
Bring the same operations to your terminal and CI.
Read guideMake your servers feel like one system.
Start with a single manager. Grow your cluster when you need to.