Plugin System

Nuvom v0.9 introduced first-class plugin support — allowing developers to register their own backends, extensions, and instrumentation layers with clean lifecycle management.

---

What is a Plugin?

A plugin is a Python class that implements the Plugin protocol, defining:

• name: Unique plugin name
• provides: What kind of capability it offers (e.g. queue_backend, monitoring)
• start(settings: dict): Initialization logic
• stop(): Cleanup logic on shutdown

---

File Layout

Plugins can live inside your project or in external libraries. Common layout:

my_project/
├── plugins/
│   ├── my_plugin.py       # Your plugin logic
│   └── __init__.py
└── .nuvom_plugins.toml    # Plugin registration

---

Plugin Example: Custom Queue Backend

# plugins/my_queue.py
from nuvom.queue_backends.base import BaseJobQueue
from nuvom.plugins.contracts import Plugin

class MyQueue(BaseJobQueue):
    def enqueue(self, job): ...
    def dequeue(self, timeout=None): ...
    def pop_batch(self, n): ...
    def qsize(self): ...
    def clear(self): ...

class MyPlugin(Plugin):
    name = "my_queue"
    provides = ["queue_backend"]

    def start(self, settings):
        from nuvom.plugins.registry import register_queue_backend
        register_queue_backend("my_custom_queue", MyQueue)

    def stop(self):
        pass

Then register in .nuvom_plugins.toml:

[plugins]
queue_backend = ["plugins.my_queue:MyPlugin"]

---

Plugin Registry

Nuvom loads plugins at worker startup (not during CLI commands).

Each plugin is discovered from .nuvom_plugins.toml, instantiated, and injected with runtime settings.

To test plugin loading:

nuvom plugin test
nuvom plugin list
nuvom plugin inspect my_queue

---

Plugin Lifecycle

Each plugin receives a start(settings) call when the worker starts.

Example use cases:

• Registering new backends
• Spawning monitoring threads
• Configuring metrics exporters
• Connecting to databases

On shutdown (CTRL+C, SIGINT), plugins receive a stop() call to clean up.

---

Plugin Types

Provides	Purpose
queue_backend	Add a new job queue implementation
result_backend	Add a result/error store
monitoring	Expose runtime metrics (e.g. Prometheus)
lifecycle_hook	Run logic on job events (future)

---

Monitoring Plugin Example

# plugins/prometheus_exporter.py
from nuvom.plugins.contracts import Plugin
from nuvom.plugins.registry import register_metrics_provider

class PrometheusPlugin(Plugin):
    name = "prometheus"
    provides = ["monitoring"]

    def start(self, settings):
        from my_exporter import run_exporter
        run_exporter(port=settings.get("prometheus_port", 9150))

    def stop(self):
        ...

---

Best Practices

• Keep plugins self-contained.
• Use settings from .env or custom sources.
• Fail fast if required dependencies are missing.
• Avoid blocking inside start() — spawn threads if needed.
• Don’t forget stop() for graceful cleanup.

---

CLI Support

Command	Description
nuvom plugin test	Validates and loads plugins
nuvom plugin list	Lists registered plugin types
nuvom plugin inspect	Shows metadata for one plugin

---

Roadmap (Future Plugin Hooks)

Feature	Status
before_job, after_job	[ ] Planned (via plugin)
metrics_provider interface	✅ Prometheus supported
CLI plugin extensions	[ ] Future
Multi-process plugin runners	[ ] Future

---

Summary

Plugins allow you to:

• Add new backends (queue, result)
• Hook into Nuvom’s lifecycle
• Export runtime metrics
• Customize behavior without modifying Nuvom core

Plugins make Nuvom truly pluggable. Add what you need — nothing more.