loki/docs/sources/installation/simple-scalable-docker.md

title	weight
Simple scalable cluster	35

Install and deploy a simple scalable cluster with Docker compose

A local Docker compose installation of Grafana Loki and Promtail is appropriate for an evaluation, testing, or development environment. Use a Tanka or Helm process for a production environment.

This installation runs Loki in a simple scalable deployment mode with one read path component and one write path component.

Prerequisites

• Docker
• Docker Compose

Obtain Loki and Promtail configuration files

Download loki-config.yaml, promtail-config.yaml, and docker-compose.yaml to your current directory:

wget https://raw.githubusercontent.com/grafana/loki/main/production/simple-scalable/promtail-config.yaml -O promtail-config.yaml
wget https://raw.githubusercontent.com/grafana/loki/main/production/simple-scalable/loki-config.yaml -O loki-config.yaml
wget https://raw.githubusercontent.com/grafana/loki/main/production/simple-scalable/docker-compose.yaml -O docker-compose.yaml

This will download loki-config.yaml, promtail-config.yaml, and docker-compose.yaml to your current directory.

The docker-compose.yaml relies on the Loki docker driver, aliased to loki-compose, to send logs to the loki cluster. If this driver is not installed on your system, you can install it by running the following:

docker plugin install grafana/loki-docker-driver:latest --alias loki-compose --grant-all-permissions

If this driver is already installed, but under a different alias, you will have to change docker-compose.yaml to use the correct alias.

Deploy and verify readiness of the Loki cluster

From the directory containing the configuration files, deploy the cluster with docker-compose:

docker-compose up

The running Docker containers use the directory's configuration files.

Navigate to http://localhost:3100/ready to check for cluster readiness. Navigate to http://localhost:3100/metrics to view the cluster metrics. Navigate to http://localhost:3000 for the Grafana instance that has Loki configured as a datasource.

By default, the image runs processes as user loki with UID 10001 and GID 10001. You can use a different user, specially if you are using bind mounts, by specifying the UID with a docker run command and using --user=UID with numeric UID suited to your needs.