Enterprise-grade monitoring and metrics collection for Dhruva Platform deployments

The observe-util package provides comprehensive observability, monitoring, and metrics tracking for FastAPI applications. It enables multi-tenant metrics collection with organization-level isolation, automatic service detection, and real-time analytics through Prometheus and Grafana.

This package is designed for the Dhruva Platform and provides plug-and-play monitoring capabilities.

• Multi-tenant metrics collection with organization-level isolation
• Automatic service detection (Translation-NMT, Text to Speech-TTS, Speech to Text-ASR, Image to Text-OCR, Named Entity Recognition-NER, Transliteration, Text Language detection, Speaker Diarization, Language Diarization, Audio Language Detection)
• Real-time monitoring via Prometheus and Grafana
• Metrics Configuration: Three built-in metrics for Latency, Error rate, Payload Size and the ability to configure customised metrics
• Business metrics - Characters, Tokens, Audio Length, Image size processed
• Easy integration - Few lines of code to get started
• Plug-and-play - Works with any FastAPI application
• Grafana Dashboards: Pre-configured dashboard templates

Get a visual overview of the monitoring capabilities provided by observe-util. The pre-configured Grafana dashboards offer comprehensive insights into your application's performance, errors, and resource usage.

Track the volume of data processed by each service with real-time payload size metrics.

Monitor service health with real-time error rate tracking and detailed breakdowns by error type. The dashboard provides percentage-based error rates and visualizations showing the distribution of different error types.

Visualize service performance with color-coded gauge charts displaying latency metrics for all services. The gauges provide instant visual feedback with green (good), orange (warning), and red (critical) performance thresholds.

The observe-util package exposes metrics in Prometheus format, enabling detailed querying and analysis of your application's telemetry data. All metrics are collected with multi-tenant support, allowing you to filter and analyze data by organization, service, endpoint, and more.

Query the telemetry_obsv_requests_total metric to analyze request patterns, view request counts across different endpoints, organizations, and HTTP status codes. This enables detailed analysis of API usage patterns and service health.

Please check https://github.com/COSS-India/observe-util/wiki for detailed documentation