Monitoring Dashboard

Rustvello ships a built-in web-based monitoring dashboard in the rustvello-monitoring crate. It provides real-time visibility into invocations, runners, workers, workflows, and task timelines — with no external dependencies beyond the Axum web server.

---

Features

SVG Timeline

Visualizes invocation scheduling across runners and workers. Each runner lane shows concurrent sub-lanes when tasks overlap, coloured by status (pending = orange, running = green, success = grey). Workers appear as distinct sub-lanes under their parent runner.

Log Explorer

Full-text search across structured logs. Cross-highlights related runners, workers, and invocations when hovering a chip. Supports filtering by runner ID, worker ID, invocation ID, task key, log level, and time range.

Invocation Tables

Filterable tables for every invocation status. Drill into individual invocations to see the full status history, call arguments, result, error, runner assignment, and workflow parent/child relationships.

Workflow View

Parent-child invocation trees rendered as interactive family graphs. Useful for debugging fan-out patterns and understanding orchestration chains.

Prometheus Metrics

When rustvello-prometheus is active, the dashboard exposes /metrics in Prometheus text format. Counters and histograms cover invocation counts, status transitions, broker queue depth, and runner heartbeat timings.

Multi-App Support

A single monitoring server can observe multiple RustvelloApp instances. Switch between apps from the navigation bar without restarting.

---

Running the Dashboard

The monitoring server is started programmatically alongside your application:

use std::collections::HashMap;
use rustvello_monitoring::{start_monitor, AppInstance, MonitorConfig};
use std::net::SocketAddr;

// Build your app normally
let app = Rustvello::builder()
    .app_id("my-app")
    .auto_discover_tasks()
    .build().await?;

// Wrap it in an AppInstance for the monitoring server
let instance = AppInstance {
    app_id: "my-app".into(),
    config: app.config.clone(),
    broker: app.broker(),
    orchestrator: app.orchestrator(),
    state_backend: app.state_backend(),
    client_data_store: app.client_data_store(),
    task_ids: app.registered_task_ids(),
};

// Start the dashboard (runs until the process exits)
start_monitor(
    HashMap::from([("my-app".into(), instance)]),
    "my-app",
    MonitorConfig {
        bind: SocketAddr::from(([0, 0, 0, 0], 8000)),
        log_level: "info".into(),
    },
).await?;

Open a browser at http://localhost:8000 to view the dashboard.

---

MonitorConfig

Field	Type	Default	Description
bind	SocketAddr	127.0.0.1:8000	Address and port to listen on
log_level	String	"info"	Monitoring server log level setting

---

Adding to Cargo.toml

[dependencies]
rustvello-monitoring = "0.1.0"

The monitoring crate does not require rustvello’s feature flags — it depends directly on rustvello-core traits and works with any backend combination.

---

Prometheus Metrics

Enable the prometheus feature on rustvello and add the rustvello-prometheus dependency to wire Prometheus metrics alongside the dashboard:

[dependencies]
rustvello = { version = "0.1.0", features = ["prometheus"] }
rustvello-prometheus = "0.1.0"
rustvello-monitoring = "0.1.0"

The dashboard automatically serves the /metrics endpoint when the Prometheus EventEmitter is registered with the app.

Metrics exported include:

Metric	Type	Description
rustvello_invocations_total	Counter	Invocations registered by task and app
rustvello_status_transitions_total	Counter	Status transitions by from/to state
rustvello_task_duration_seconds	Histogram	Task execution time by task ID
rustvello_broker_queue_depth	Gauge	Current number of pending invocations
rustvello_runner_heartbeats_total	Counter	Heartbeat events per runner

---

Structured Logging

Rustvello introduces a unified logging format that produces consistent output from both the Rust engine and Python task code. Text logs carry [R] or [P] system tags; JSON logs use "system": "rust" or "system": "python".

The dashboard’s log explorer reads these structured logs for cross-entity highlighting.

For the full logging guide — unified format specification, JSON output schema, configuration, and integration with ELK/Datadog/Loki — see Logging Guide.

Quick Setup

For the Rust CLI:

RUST_LOG=info rustvello run --app-id my-app

For Python/pynenc:

PYNENC__LOGGING_LEVEL=info PYNENC__LOG_FORMAT=json pynenc --app=tasks.app runner start

The log explorer parses runtime context (runner_id, worker_id, invocation_id, task_id) from either unified text or JSON logs.

---

Security Notes

The monitoring dashboard is intended for internal/trusted networks. It does not implement authentication by default. For production deployments:

• Bind to 127.0.0.1 and expose via a reverse proxy (nginx, Caddy) with authentication

• Or restrict network access to the monitoring port at the firewall/VPC level

CSRF origin validation is enforced on all state-mutating routes.