resource-reporter 0.1.2

A library for monitoring and reporting resource usage in containerized applications

Resource Reporter

A Python library for monitoring and reporting resource usage in containerized applications, with support for asynchronous event reporting and command listening.

Features

• Regular resource usage reporting (CPU, memory, network, GPU)
• Asynchronous event reporting
• Command listening via Google PubSub for remote control of containers
• Support for graceful shutdown and restart

Installation

Basic Installation

pip install resource-reporter

With GPU Support

pip install resource-reporter[gpu]

Quick Start

from resource_reporter import ResourceReporter

# Initialize the reporter
reporter = ResourceReporter(
    metrics_endpoint_url="http://monitoring-service/api/metrics",
    events_endpoint_url="http://monitoring-service/api/events",
    interval=30,
    consumer_id="consumer-1",
    group_consumer="production"
)

# Start reporting
reporter.start()

# Report an event
reporter.async_report_event({
    "status": 200,
    "error_code": 0,
    "error_message": "",
    "result": {"processed": True},
    "stats_data": {"duration_ms": 120},
    "task_id": "task-123"
})

# Update job stats
reporter.update_job_stats({
    "processed": 1000,
    "errors": 5
})

# Stop the reporter when done
reporter.stop()

Documentation

Environment Variables

The Resource Reporter library uses several environment variables for configuration. While most parameters can be passed directly to the constructor, environment variables provide a convenient way to configure the reporter, especially in containerized environments.

Required Environment Variables

None of the environment variables are strictly required, as all configuration can be passed directly to the constructor. However, the following are commonly used:

Optional Environment Variables

Variable	Description	Default
GROUP_CONSUMER	Group identifier for the consumer	"default"
API_KEY	API key for authentication with metrics and events endpoints	None
GOOGLE_APPLICATION_CREDENTIALS	Path to Google Cloud credentials JSON file	None

Google Cloud Authentication

For PubSub command listening functionality, the reporter requires Google Cloud authentication. There are several ways to provide this:

1. Service Account JSON (recommended for non-GCP environments):

   export GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account-key.json
   

License MIT