Router Cluster Management | Cosmo by WunderGraph - WunderGraph
See every running router from a single dashboard
Cosmo Studio automatically discovers all router instances via built-in OpenTelemetry instrumentation. Monitor CPU, memory, version, and deployed graph composition across your entire fleet. No extra setup required.
Requires router version 0.66.1 or later.
The problem
Operating routers in production without fleet-level visibility
As router deployments grow across environments and regions, tracking versions, health, and composition across the fleet requires tooling that generic monitoring cannot provide.
Operating a router fleet without visibility
Multiple router instances across environments make it hard to answer basic questions: how many are running, which versions are deployed, and whether any are under resource pressure.
Deployment verification is a manual process
After rolling out a new router version, confirming that every instance updated requires checking each one individually. A centralized view would make this instant.
Composition mismatches go undetected
Different router instances may run different graph compositions during a rollout. Without visibility into which composition each router runs, inconsistencies are hard to spot.
Our solution
Fleet visibility with zero setup
Routers automatically report telemetry to Cosmo Cloud. Studio aggregates this data into a real-time dashboard showing every running instance across all clusters.
How the dashboard works
- Deploy routers with version 0.66.1 or later.
- Routers automatically send periodic telemetry to Cosmo Cloud via built-in OpenTelemetry instrumentation.
- Cosmo Studio aggregates this data and displays all running router instances in the Cluster Management dashboard.
- Each instance shows name, version, cluster, uptime, and resource utilization.
- An instance is considered offline if it fails to report within 45 seconds.
- Group instances logically using the CLUSTER_NAME environment variable.
Before & After
| Before | With Cluster Management |
|---|---|
| Manual tracking of which router instances are running | Automatic discovery and display in Studio |
| Monitoring scattered across separate tools and dashboards | Centralized view of the entire router fleet in Studio |
| Unknown versions deployed across the fleet | Clear version visibility per instance |
| Resource issues discovered after user impact | Real-time CPU and memory with trend indicators |
What the dashboard shows
- Name: TELEMETRY_SERVICE_NAME (default: cosmo-router)
- Instance ID: INSTANCE_ID env var, auto-generated if not set
- Status: Up/running indicator; instance drops if no sample in 45s
- Version: Deployed router binary version
- Cluster: Logical group via CLUSTER_NAME env var
- Uptime: Process and graph composition uptime
- Mem / CPU: Utilization with trend arrows
How Router Cluster Management works
Auto-discover
Routers with version 0.66.1 or later automatically send periodic telemetry to Cosmo Cloud via built-in OpenTelemetry instrumentation. No additional configuration required.
Monitor
The Cluster Management dashboard in Studio displays CPU and memory utilization for each instance. Trend arrows show whether utilization is increasing or decreasing between samples.
Verify
Click any router instance to see its deployed binary version and the graph composition it currently runs. Confirm that rollouts completed as expected across the fleet.
Group
Set the CLUSTER_NAME environment variable on your routers to organize instances by logical cluster (e.g., production-us, staging-eu). View and filter by cluster in the dashboard.
Operational use cases
Deployment verification
After rolling out a new router version, open the dashboard, filter by cluster, and confirm every instance shows the new version. No SSH required.
Resource pressure detection
Monitor CPU and memory utilization across the fleet. Trend arrows signal which instances are under growing pressure before it affects request latency.
Composition verification
Confirm which graph composition each router instance runs during a rollout. Catch situations where an instance still runs an outdated composition.
Multi-cluster overview
Set CLUSTER_NAME on each router group (production-us, staging-eu, etc.) to view and manage all environments from a single dashboard in Studio.
Get visibility into your router fleet
Automatic discovery. Real-time vitals. Requires router v0.66.1+. Available on Pro and Enterprise.
FAQ
Cluster Management questions
- What router version is required for Cluster Management?
- Do I need to configure anything extra to enable the dashboard?
- How long before an offline router disappears from the dashboard?
- How do I give a router a stable identity in the dashboard?
- How do I group routers by environment or region?
- What plan is required for Cluster Management?
- Can I view the graph composition a specific router is running?
Read more in the Cluster Management documentation.