Lab: Deploy Loki and send your first logs

What you will learn in this TP

• Start Loki and Grafana
• Push and read a log through the API
• Use logcli

Introduction #

What you will learn in this section

• Start Loki and Grafana
• Push and read a log through the API
• Use logcli

Context. Your team stores its logs in files on each server: it is impossible to view them together or to correlate them with metrics. You install Loki, the “Prometheus-style” log system, to understand how it receives and returns a log.

Prerequisites: Docker and Docker Compose  ·  Estimated time: 30 min

You will deploy a local Loki + Grafana stack and validate each building block with commands.

Setup #

Configuration, startup and validation.

1. 1. Create loki.yaml and docker-compose.yml

   Why? A reproducible local stack lets you test Loki without touching production; the configuration stores everything on local disk to keep things simple.

   Reuse the monolithic configuration from the installation article (filesystem storage, TSDB schema v13) and declare the loki and grafana services.

2. 2. Check that Loki is ready

   Why? The /ready endpoint tells you Loki has finished starting up: sending logs before that point causes misleading errors.

   Which command shows that Loki is accepting requests?

3. 3. Push a log with the API

   Why? The push API is what every agent (Alloy, Promtail) will use: handling it by hand shows what they actually send (labels + timestamp + line).

   Send the line “Hello Loki” to the {job="tp"} stream.

4. 4. Read with logcli and Grafana

   Why? logcli and Grafana are the two ways to view logs; knowing both means you do not depend on a single interface when troubleshooting.

   Find the log with logcli, then add the Loki data source in Grafana (http://loki:3100) and run the query in Explore.

Difficulty level : ●○○○○ (1/5)