feat: add Prometheus datasource connector for real-time metrics querying

[RichardoMrMu]

Summary

Add a PrometheusConnector that enables OpenDerisk agents to query Prometheus HTTP API for real-time metrics data, replacing the dependency on static OpenRCA datasets for metrics analysis.

Motivation

Currently, OpenDerisk's SRE agents rely on static OpenRCA datasets (CSV files) for metrics analysis during root cause diagnosis. This limits the system to offline/historical analysis and prevents real-time monitoring and diagnostics.

By adding a Prometheus datasource connector, agents can:

• Query live production metrics via PromQL
• Perform real-time anomaly detection using actual Prometheus data
• Access scrape targets, alerting rules, and active alerts for comprehensive diagnostics
• Integrate with existing Prometheus-based monitoring infrastructure

Changes

Action	File	Lines
Added	packages/derisk-ext/src/derisk_ext/datasource/conn_prometheus.py	+460
Added	packages/derisk-ext/src/derisk_ext/datasource/tests/__init__.py	+0
Added	packages/derisk-ext/src/derisk_ext/datasource/tests/test_conn_prometheus.py	+229

Key Features

• Instant query — PromQL evaluation at a single point in time
• Range query — PromQL evaluation over a time range with configurable step
• Series discovery — Find time series matching label selectors
• Label enumeration — List label names and values
• Target and rule inspection — Scrape targets, alerting/recording rules
• Active alerts listing — Current firing alerts
• Health check — Prometheus server health verification
• Metric metadata — Type, help text, and unit information

Design Decisions

• Inherits from BaseConnector (not RDBMSConnector) since Prometheus is a time-series database, not a relational database
• Uses requests library for HTTP API calls (already a project dependency)
• Implements run() method for compatibility with the BaseConnector interface, treating the command as a PromQL expression
• Supports authentication via basic auth and custom headers
• SSL/TLS configurable for production environments

Usage Example

from derisk_ext.datasource.conn_prometheus import PrometheusConnector

connector = PrometheusConnector(
    host="prometheus.example.com",
    port=9090,
    scheme="https",
)

# Instant query
results = connector.instant_query('up{job="derisk"}')

# Range query
results = connector.range_query(
    query='rate(http_requests_total[5m])',
    start='2024-01-01T00:00:00Z',
    end='2024-01-01T01:00:00Z',
    step='60s',
)

# Check health
is_healthy = connector.check_health()

# Get active alerts
alerts = connector.get_alerts()

Testing

Comprehensive unit tests with mocked HTTP responses covering:

• Parameter configuration and URL construction
• Instant and range query execution
• run() interface compatibility
• Health check success/failure
• Error handling (API errors, connection failures)
• Result formatting for both instant and range query results

Related

• Complements PR [Feature] Add Prometheus metrics endpoint for OpenDerisk observability #128 (Prometheus metrics export) by adding the read side
• Part of the observability enhancement roadmap discussed in Issue [Discussion] OpenTelemetry Tracing Enhancement - Semantic Conventions, SpanKind, and Production-Grade Features #172