Installation on Linux

If you ever get stuck during this installation, be sure to reboot the machine once. It may help to correctly load some configurations and/or daemons.

To get correct measurements, the tool requires a linux distribution as foundation, a webserver (instructions only given for NGINX, but any webserver will do), python3 including some packages, and docker installed (rootless optional). In this manual we are assuming you are running a Debian/ Ubuntu flavour of Linux.

Currently the following distributions have been tested and are fully supported:

• Ubuntu 22.04
• Fedora 38

The following distributions have been tested, but require manual work:

• Ubuntu 20.04 (works, but libglib has to be manually updated to libglib2.0-dev)
• Ubuntu 22.10 (works for development, but cluster installation has different names for timers)

Downloading and installing required packages #

For the sake of this manual we put the green metrics tool into your home directory. Of course you can place it anywhere. Also we trigger a apt-upgrade. If you do not want that upgrade or a different path for the tool please modify the commands accordingly.

• Ubuntu
• Fedora

sudo apt update && \
sudo apt upgrade -y && \
sudo apt install -y curl git make gcc python3 python3-pip python3-venv && \
git clone https://github.com/green-coding-berlin/green-metrics-tool ~/green-metrics-tool && \
cd ~/green-metrics-tool

sudo dnf upgrade -y && \
sudo dnf install -y curl git make gcc python3 python3-devel python3-venv && \
git clone https://github.com/green-coding-berlin/green-metrics-tool ~/green-metrics-tool && \
cd ~/green-metrics-tool

Docker #

Docker provides a great installation help on their website that will probably be more up to date than this readme: https://docs.docker.com/engine/install/

This is also the reason why the docker client is not part of the install script that we provide.

However, we provide here what we used in on our systems, but be sure to double check on the official website. Especially if you are running a different distribution.

Docker base install #

• Ubuntu
• Fedora

sudo apt install ca-certificates curl gnupg lsb-release -y && \
sudo install -m 0755 -d /etc/apt/keyrings && \
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg && \
sudo chmod a+r /etc/apt/keyrings/docker.gpg && \
echo "deb [arch="$(dpkg --print-architecture)" signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
  "$(. /etc/os-release && echo "$VERSION_CODENAME")" stable" | \
  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null && \
sudo apt update && \
sudo apt remove docker docker.io docker-doc docker-compose docker-compose-v2 podman-docker containerd runc -y && \
sudo apt install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin -y

sudo dnf remove docker \
                  docker-client \
                  docker-client-latest \
                  docker-common \
                  docker-latest \
                  docker-latest-logrotate \
                  docker-logrotate \
                  docker-selinux \
                  docker-engine-selinux \
                  docker-engine
sudo dnf -y install dnf-plugins-core
sudo dnf config-manager \
    --add-repo \
    https://download.docker.com/linux/fedora/docker-ce.repo
sudo dnf install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
sudo systemctl start docker

You can check if everything is working fine by running docker stats. It should connect to the docker daemon and output a view with container-id, name, and stats, which should all be empty for now. You can also run sudo docker run hello-world which will run a little welcome container.

Root mode (default) #

Since v0.20 the Green Metrics Tool supports docker in it’s default root mode.

However you need to add your current user to the docker group. We need this so we do not need to start the Green Metrics Tool with root privileges.

Please follow this explanation how to do it: Official docker docs on docker group add

=> If you want to use the rootless mode anyway you do not have to to that. Just read the next paragraph.

Rootless mode (optional) #

Rootless mode allows the docker container to not inherit root rights when they run.

In order to use rootless mode you must have a non-root user on your system (see https://docs.docker.com/engine/security/rootless/

👉 Typically a normal installation of Ubuntu/ Fedora has at least one non-root user setup during installation, so there is no need to create a different user from the one you are using already.

Important: If you have just created a non root user be sure to relog into your system (either through relogging, or a new ssh login) with the non-root user. A switch with just su my_user will not work.

• Ubuntu
• Fedora

sudo systemctl disable --now docker.service docker.socket && \
sudo apt install uidmap && \
sudo apt update && \
sudo apt-get install -y docker-ce-rootless-extras dbus-user-session && \
dockerd-rootless-setuptool.sh install

systemctl --user enable docker
sudo loginctl enable-linger $(whoami)

sudo systemctl disable --now docker.service docker.socket && \
sudo dnf install -y shadow-utils fuse-overlayfs iptables && \
sudo dnf install -y docker-ce-rootless-extras && \
dockerd-rootless-setuptool.sh install

systemctl --user enable docker
sudo loginctl enable-linger $(whoami)

You must also enable the cgroup2 support with the metrics granted for the user: https://rootlesscontaine.rs/getting-started/common/cgroup2/.

Make sure to also enable the CPU, CPUSET, and I/O delegation as instructed there.

Setup #

Please run the install_linux.sh script in the root folder.

This script will:

• Ask for the URLs of where to deploy the frontend and API
  • If you are working locally we strongly encourage you to use the defaults of http://metrics.green-coding.internal:9142 and http://api.green-coding.internal:9142. All other local domains are not supported out of the box.
  • If you plan to deploy on an outside visible URL please type the URL including https:// but omitting port if it is running on port 80 or 443
• Set the database password for the containers
  • By default the script will ask you to provide a password, but you can also pass it in directly with the -p parameter.
• Initialize and update git submodules
• Install a python venv and activate it
• Create the needed /etc/hosts entries for development
• Install needed development libraries via apt for metric providers to build
• Build the binaries for the Metric Providers
• Set needed /etc/sudoers entry for requesting kernel scheduler info

Please note that whenever you run the Green Metrics Tool you have to first activate the python venv.

What you might want to add:

• SMTP mail sending is by default deactivated, so for a quick-start you do not have to change that in the config.yml
• The RAPL reporter is by default deactivated. Please check the Metric Providers Documentation on how to active it

After that you can start the containers (use sudo if running in docker root mode):

• Build and run in the docker directory with [sudo] docker compose up
• The compose file uses volumes to persist the state of the database even between rebuilds. If you want a fresh start use: [sudo] docker compose down -v && [sudo] docker compose up
• To start in detached mode just use [sudo] docker compose up -d

Connecting to DB #

You can now connect to the db directly on port 9573, which is exposed to your host system.
This exposure is not strictly needed for the green metrics tool to run, but is useful if you want to access the db directly. If you do not wish to do so, just remove the 9573:9573 entry in the compose.yml file.

The database name is green-coding, user is postgres, and the password is what you have specified during the install_linux.sh run, and can be found in the compose.yml file.

Restarting Docker containers on system reboot #

We recommend systemd. Please use the following service file and change the USERNAME accordingly to the ones on your system.

The file will be installed to: /home/USERNAME/.config/systemd/user/green-coding-service.service

[Unit]
Description=Docker Compose for all our services
After=docker.service

[Service]
Type=simple
ExecStart=/home/USERNAME/startup-docker.sh
Restart=on-failure
RestartSec=5s
StartLimitBurst=10
StartLimitInterval=0

[Install]
WantedBy=default.target

As you can see Restart is set to on-failure. The reason is that the docker dameon will restart the containers by itself. The systemd script is only needed to start the container once on reboot.

As you can see we also reference the /home/USERNAME/startup-docker.sh file which systemd expects to be in your green metrics tool directory.

Please create the following file in your home directory:

#!/bin/bash
docker context use rootless
docker compose -f /home/USERNAME/docker/compose.yml up -d

Now you can reload and enable the daemon:

systemctl --user daemon-reload
systemctl --user enable green-coding-service

Since the service runs as a user it might be the case that the user is not logged in. So we also enable lingering which guarantees that processes can be started when no user is logged in:

loginctl enable-linger $(whoami)

Dockerfiles architecture explanation: #

• The postgres container has a volume mount. This means that data in the database will persist between container removals / restarts
• The interconnect between the gunicorn and the nginx container runs through a shared volume mount in the filesystem. Both use the user www-data to read and write to a UNIX socket in /tmp
• all webserver configuration files are mounted on start of the container as read-only. This allows for changing configuration of the server through git-pull or manual editing without having to rebuild the docker image.
• postgresql can detect changes to the structure.sql. If you issue a docker compose down -v the attached volume will be cleared and the postgres container will import the database structure fresh.

Metric providers #

Some metric providers need extra setup before they work.

LM-Sensors #

The required libraries are installed automatically via the install-linux.sh call. Some modifications need to be made to your config.yml file though, which are detailed in the lm-sensors metric provider documentation →, along with further details regarding prerequisites.

XGBoost #

The XGBoost metrics provider can estimate the power consumption of the total system (AC-Energy). It is included as a submodule in the Green Metrics Tool and should have been checked out with the initial install command of this manual. The config.yml file also needs additional details which are detailed in the metric provider documentation→.

NVIDIA SMI #

The NVIDIA SMI metrics reporter can read the power draw of an NVIDIA GPU. Some libraries need to be installed in order for the NVIDIA SMI metric provider to work.

DC Metrics Provider #

Some of our PSU metrics providers may need specific hardware attached to your machine in order to run. These include the Powerspy, MCP, and Picolog metric providers. Please look for details in each provider’s corresponding documentation

RAPL #

On kernels > 2.6 all the kernel modules should automatically be loaded.

However just in case run:

sudo modprobe intel_rapl_common # or intel_rapl for kernels < 5
sudo modprobe intel_rapl_msr
sudo modprobe rapl

Live system #

ℹ️ If you just want to run the Green Metrics Tool locally this step can be skipped
ℹ️ It is only if you want to host the Green Metrics Tool on a live server.

Updating port to 80 #

The development setup of the GMT binds on port 9142. For a normal setup on a live server we recommend binding it to port 80.

The change is done in the /docker/compose.yml file.

green-coding-nginx:
    [...]
    ports:
      - 9142:80 # change this to 80:80

SSL #

TODO!

No thorough documentation on this yet! However you have to configure NGINX accordingly so that it finds the SSL credentials and certificate. This is done in the /docker/nginx/frontend.conf.