Environment:Evidentlyai Evidently Grafana Monitoring Environment

Knowledge Sources	Evidently Data Drift Grafana Example
Domains	Infrastructure, Monitoring
Last Updated	2026-02-14 10:00 GMT

Overview

Docker Compose environment with PostgreSQL, Grafana, and Adminer for visualizing Evidently metrics on real-time monitoring dashboards.

Description

This environment provides a complete monitoring stack for visualizing Evidently data drift and LLM evaluation metrics. It uses Docker Compose to orchestrate PostgreSQL (for metric storage), Grafana Enterprise (for dashboards), and Adminer (for database inspection). Evidently metrics are calculated in a Python script and inserted into PostgreSQL, then visualized through preconfigured Grafana dashboards.

Usage

Use this environment for production monitoring dashboards that visualize data drift or LLM evaluation metrics over time. Required for the `Evidentlyai_Evidently_Data_Drift_Monitoring` and `Evidentlyai_Evidently_LLM_Evaluation_Monitoring` workflows.

System Requirements

Category	Requirement	Notes
OS	Linux, macOS	Docker-compatible OS
Docker	Docker Engine + Docker Compose	v3.7+ compose file format
Ports	5432, 8080, 3000	PostgreSQL, Adminer, Grafana respectively
Disk	~1GB	For Docker images and data volumes

Dependencies

System Packages

Docker Engine
Docker Compose (v3.7+ file format)

Docker Images

`postgres` (latest or `postgres:16-alpine`)
`grafana/grafana-enterprise`
`adminer`

Python Packages (for metrics calculation script)

`prefect` (workflow orchestration, data drift example)
`tqdm` (progress bars)
`psycopg` / `psycopg_binary` (PostgreSQL driver)
`pyarrow` (Parquet support)
`evidently` (core library)
`pandas`, `numpy`, `scikit-learn` (data processing)
`matplotlib` (optional visualization)
`openai` (LLM evaluation example only)

Credentials

The following credentials are configured in `docker-compose.yml`:

`POSTGRES_PASSWORD`: PostgreSQL password (set in Docker Compose environment).
`POSTGRES_USER`: PostgreSQL username (for service example: `evidently_user`).
`POSTGRES_DB`: PostgreSQL database name (for service example: `evidently`).

Quick Install

# Clone and start the monitoring stack
cd examples/data_drift_grafana_dashboard
docker compose up -d

# Install Python dependencies for metrics calculation
pip install -r requirements.txt

Code Evidence

Docker Compose service definitions from `examples/data_drift_grafana_dashboard/docker-compose.yml:11-42`:

services:
  db:
    image: postgres
    restart: always
    environment:
      POSTGRES_PASSWORD: example
    ports:
      - "5432:5432"

  adminer:
    image: adminer
    restart: always
    ports:
      - "8080:8080"

  grafana:
    image: grafana/grafana-enterprise
    user: "472"
    ports:
      - "3000:3000"

Service Docker Compose with health checks from `examples/service/docker-compose.yml:21-40`:

evidently-service:
    image: evidently/evidently-service:latest
    depends_on:
      postgres:
        condition: service_healthy
    command: ["--port", "8000", "--conf-path", "/app/config.yaml"]
    environment:
      EVIDENTLY_DEBUG: 1

Common Errors

Error Message	Cause	Solution
Port 5432 already in use	Local PostgreSQL running on same port	Stop local PostgreSQL or change the port mapping in docker-compose.yml
Port 3000 already in use	Another service on Grafana port	Change the Grafana port mapping in docker-compose.yml
Grafana datasource connection refused	PostgreSQL not ready when Grafana starts	Docker Compose uses `restart: always`; wait for PostgreSQL to be ready

Compatibility Notes

Docker Compose versions: The data drift example uses v3.7 format; the service example uses v3.8 format.
Grafana user: The Grafana container runs as user "472" for correct permissions on provisioned dashboards.

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment