Valkey Mesos Control Plane

Dual implementation reference:

• backend/app.go: Process start, HTTP server and lifecycle.
• backend/init.go: Central initialization of all environment variables via init().
• backend/main.go: API, domain models and Mesos abstraction.
• frontend/: React/Vite administration frontend with live polling.

Start

make run

This starts the backend on http://localhost:8080 and frontend on http://localhost:5173. For a secure local test, the backend scheduler runs by default in DRY_RUN=true; nodes and scaling are then deterministically visible, but not actually started on Mesos.

Dashboard

The React dashboard provides a live cluster overview, node resource details, scheduler events, and explicit start/stop and scaling controls.

For the test cluster, credentials can only be set via shell:

MESOS_MASTER=mesos.example.test:5050 MESOS_USERNAME=example-user MESOS_PASSWORD="$MESOS_PASSWORD" DRY_RUN=true make run

DRY_RUN=false activates the explicit production seam. The Mesos-v1 framework registration/streaming integration is intentionally not yet activated; the backend startup will fail with an explanatory error message on scaling instead of pretending to have a successful deployment status.

API

• GET /api/health
• GET /api/cluster
• PUT /api/cluster/scale with { "nodes": 3..100 }
• GET /api/metrics

The metrics explicitly report unknown_no_valkey_endpoint in dry-run mode because no real Valkey processes are running. Once Mesos task status and endpoints are integrated, the backend metrics collector will query Valkey INFO via TCP.

Foundation

The scheduler structure is based on m3scluster/compose: Framework/Mesos client behind an interface, reconciliation as source for desired task count, and offer-/task-oriented scheduling. The external integration remains separate from the local dry-run implementation.