🚀 Container Exporter (CXP)

A resource-friendly, highly efficient, and minimal Prometheus exporter to track Memory, CPU, Disk and Network I/O usage of Docker containers along with their lifecycle (up/down) state used for alerting.

Check out the web page for more information.

Table of Contents

1. DEV STACK
2. DEMO
3. Step-by-Step Guide
   1. Before You start
   2. Getting started
      • Deploy with Github Actions
      • Deploy with Docker
      • Deploy without Docker
      • Run with a Custom Port
   3. Add CXP to Prometheus
4. Grafana Dashboards
5. TO-DO
6. Contributions
7. Contact Information

🛠️ DEV STACK

see a sample of the metrics page here.

🎥 DEMO

📋 Step-by-Step Guide

Before You start

• Port 8000 must be open (default port)
• Docker & Docker Compose should be installed (optional)
• The presence of Git and Python3.10

Getting started

⚙️ Deploy with Github Actions

• fork the repository.
• go to the fork repository and switch to Action tab.
• click the I understand my workflows, go ahead and enable them button.
• now you have access to all of the workflows, however make sure you change the secrets listed below accordingly:
  1. secretes.DOCKER_TOKEN : the personal access token docker hub of your(or your organization) account.
  2. secretes.GHCR_TOKEN : github classic access token with (packages read:write permissions) or just simply use ${{github.token}}. help

here's how workflows work:

• on push to master the project will be built, deployed and released. since deploying to your servers requires runner configuration it must be triggered manually, you can modify its behavior on cd.yml workflow.
• on push to any branch except for master code will be built and healthchecked
• on pr the project will be healthchecked and built.

double check the required variables and secrets to prevent any unexpected failures

🐳 Deploy with Docker

• clone and checkout to the repository using the following commands:

  git clone https://github.com/Shayan-Ghani/Container-exporter.git
  cd Container-Exporter

• Deploy the docker-compose file that suits you the best for instance :

# Make the file executable
  chmod +x ./start.sh

# With docker compose v1
  docker-compose -f container-exporter.yml up -d
# Or using v2
  docker compose -f container-exporter.yml up -d
# build from base with Dockerfile
  docker-compose -f container-exporter-local.yml up --build -d

🐍 Can't use Docker? Ok then :

# No need if done already
  chmod +x ./start.sh

# Set up virtualenv
  python3 -m venv venv
  source venv/bin/activate
  pip install -U pip

# Install the required python packages
  pip install -r requirements.txt

# start the initializer script.
  ./start.sh &

You can use nohup as well :

nohup ./start.sh -out ./nohup.out

# to stop cxp without docker use this command 
kill -9 <PID>

Replace <PID> with the pid of ./start.sh script.

🚢 Run With A Custom Parameters:

• adjust the following settings for uvicorn as environment variables:
  • HOST (Default: 0.0.0.0)
  • PORT (Default: 8000)
  • WORKERS (Default : 1)
  • LOG_LEVEL (Default : warning)

Example:

PORT="8000" ./start.sh <your custome port> &

🔥 Add CXP to Prometheus

• Edit your prometheus.yml file and add the address of container-exporter in scrape_configs:

• Reload or restart your Prometheus server and reach out to http://127.0.0.1:8000/metrics

That is it you are good to go, Enjoy Using CXP! "}"

📊 Grafana Dashboards

Check out dashboards directory for Json files. including CPU & Memory usage + containers status (uptime).

Change Your Prometheus data source uid with the uid of Prometheus data source uid. you can find it this way:

• Reach out to Grafana then enter Home > Administration > Data sources then click on your Prometheus data source.

• the characters after datasources/edit/ are your uid. (e.g datasources/edit/c8e586ac-4262-4aad5-a103-1240ss826424)

• alternatively, use dashboard-gen.sh script to change the dashboards' uid by providing the uid as the first argument of the script. do the following steps:

 cd scripts && bash dashboard-gen.sh <your uid> 

• replace <your uid> with your Prometheus datasource uid.

• now head to Grafana dashboards and hit new > import then copy the dashboard Json file and paste it into Import via panel json

• hit the load button and Done!

TO-DO

• Disk I/O usage
• Network I/O Usage
• Add metrics in units of byte
• Check and Unregister stat metrics for containers that are not running
• Design and develop a static website to showcase Documentation, new features, etc.
• Enable functionality and smoke testing in ci
• Add clear_metrics functionality to switch on clearing the labels or setting them to 0 to maintain time series data, on user's demand.
• Design grafana dashboards and share them on grafana cloud
• Add unit tests

Contributions

Welcome to CXP! This project is production-ready now, and we encourage contributions to enhance its functionality, optimize code, and add new features

Feel free to contribute in any wacacy you can. If you come across a bug or have a suggestion, please don't hesitate to file an issue. Your input is valuable and helps us improve CXP for everyone; Therefore, add any desired function or feature to TO DO section. We appreciate your contribution to making CXP even better! If you have any questions or need assistance, feel free to reach out. Thank you!

• If you want to add metrics to cxp, make sure the naming convention is conformed to. (cxp_metric_name)