Skip to main content
Start every investigation with docker compose logs <service>. Most problems leave a clear error message in the container output that points directly to the cause.
Check the logs of the failing container first:
docker compose logs <service>
Replace <service> with the container name shown as Exited in docker compose ps — for example, prometheus, loki, promtail, node-exporter, or grafana.Common causes:
  • Port already in use. If the log contains bind: address already in use, another process on your host is occupying the same port. See the Port already in use accordion below for steps to resolve this.
  • Config file syntax error. Prometheus, Loki, and Promtail all mount their config files as volumes. A YAML syntax error in any of these files prevents the container from initialising. Check prometheus/prometheus.yml, loki/loki-config.yml, or promtail/promtail-config.yml for incorrect indentation or invalid values, then run: 
    docker compose up -d --force-recreate <service>
  • Volume path issue. If Docker cannot find a file at the path specified in the volumes: section of docker-compose.yml, the container exits immediately. Confirm that prometheus/prometheus.yml, loki/loki-config.yml, and promtail/promtail-config.yml all exist relative to the directory where you run docker compose.
Symptom: The Prometheus datasource in Grafana shows a connection error when you click Save & test.Diagnosis:
  1. Confirm Prometheus is running: 
    docker compose ps
    The prometheus container should show status Up.
  2. Check Prometheus logs for startup errors: 
    docker compose logs prometheus
Fix:Verify that the datasource URL in Grafana is set to http://prometheus:9090, not http://localhost:9090.Grafana runs inside Docker and resolves service names over the internal Docker network. Using localhost inside the container refers to the Grafana container itself, not your host machine, so the connection will always fail. The correct URL uses the Docker service name prometheus as the hostname.Go to Connections → Data sources, select your Prometheus datasource, update the URL to http://prometheus:9090, and click Save & test.
Symptom: The Loki datasource in Grafana shows a connection error, or log queries return no data.Diagnosis:
  1. Confirm Loki is running: 
    docker compose ps
    The loki container should show status Up.
  2. Check Loki logs for errors: 
    docker compose logs loki
  3. Confirm Loki is listening on port 3100 by checking its config: 
    # loki/loki-config.yml
server:
  http_listen_port: 3100
Fix:The datasource URL in Grafana must be http://loki:3100, not http://localhost:3100. Like Prometheus, Grafana resolves loki as a hostname over the internal Docker network.Go to Connections → Data sources, select your Loki datasource, update the URL to http://loki:3100, and click Save & test.
Symptom: Loki datasource connects successfully in Grafana, but log queries such as {job="varlogs"} return no results.Diagnosis:
  1. Check Promtail logs for collection errors: 
    docker compose logs promtail
    Look for messages about file discovery or push failures.
  2. Confirm the /var/log volume is mounted correctly in docker-compose.yml: 
    promtail:
  volumes:
    - ./promtail/promtail-config.yml:/etc/promtail/config.yml
    - /var/log:/var/log:ro
    If the /var/log:/var/log:ro line is missing or the path is wrong, Promtail has nothing to read.
  3. Verify the scrape path in promtail/promtail-config.yml: 
    scrape_configs:
  - job_name: system
    static_configs:
      - targets:
          - localhost
        labels:
          job: varlogs
          __path__: /var/log/*.log
    The __path__ label must match files that actually exist inside the container at /var/log/*.log. If your host system stores logs elsewhere, update the volume mount and __path__ to match.
  4. Confirm Promtail can reach Loki. Its push URL is configured as: 
    clients:
  - url: http://loki:3100/loki/api/v1/push
    If Loki is not running or the URL is wrong, Promtail will log push errors.
Fix: Correct any of the issues found above, then restart Promtail to apply the changes:
docker compose up -d --force-recreate promtail
Symptom: Metrics like node_cpu_seconds_total are missing from Prometheus, or the node-exporter target shows as DOWN on the Prometheus targets page.Diagnosis:
  1. Open the Prometheus targets page at http://localhost:9090/targets and check the status of the node-exporter job.
  2. Check Node Exporter logs: 
    docker compose logs node-exporter
  3. Confirm the scrape target in prometheus/prometheus.yml is set to node-exporter:9100: 
    scrape_configs:
  - job_name: 'node-exporter'
    static_configs:
      - targets: ['node-exporter:9100']
    Prometheus resolves node-exporter as a hostname over the internal Docker network. If this is set to localhost:9100, scraping will fail because localhost inside the Prometheus container refers to the Prometheus container itself.
  4. Confirm Node Exporter is running: 
    docker compose ps
Fix: If prometheus.yml is incorrect, update it to use node-exporter:9100 as the target, then force-recreate Prometheus to pick up the change:
docker compose up -d --force-recreate prometheus
Symptom: A container exits immediately and docker compose logs <service> contains bind: address already in use.Diagnosis:Identify which process is occupying the port. For example, to check port 9090:
sudo ss -tlnp | grep 9090
The ports used by each service are:
ServiceHost port
Grafana3000
Loki3100
Prometheus9090
Node Exporter9100
Fix:You have two options:
  • Stop the conflicting process and run docker compose up -d again.
  • Change the host port for the affected service in docker-compose.yml. The format is "<host-port>:<container-port>". For example, to move Prometheus to host port 19090: 
    prometheus:
  ports:
    - "19090:9090"
    After editing, recreate the container: 
    docker compose up -d --force-recreate prometheus
    Remember to update any URLs that reference the old port — such as the Prometheus datasource URL in Grafana or the link you use to open the Prometheus UI.