How to add LGTM data sources in Grafana

Adding LGTM data sources in Grafana connects the UI to the log, trace, and metrics backends that store telemetry. A self-managed stack usually needs a Loki data source for logs, a Tempo data source for traces, and a Prometheus-compatible data source for metrics from Mimir or Prometheus.

Provisioning keeps the data source names, backend URLs, and unique IDs in a version-controlled file instead of relying on manual UI clicks. The URLs must be reachable from the Grafana server or container, so service names such as loki, tempo, and mimir work inside a shared Docker Compose network while localhost usually points back at Grafana itself.

Use fixed data source UIDs when later guides need trace-to-log links, dashboards, alert rules, or panel JSON to reference the same backends. A saved connection should pass Save & test in the UI, and Explore should return one log stream, trace, or metric from each backend before dashboards depend on the stack.

1. Create the Grafana provisioning directory in the Compose project.

   $ mkdir -p grafana/provisioning/datasources

2. Save the LGTM data source provisioning file as lgtm.yaml in that directory.

   lgtm.yaml

   apiVersion: 1
   
   datasources:
     - name: Loki
       uid: loki
       type: loki
       access: proxy
       url: http://loki:3100
       jsonData:
         maxLines: 1000
   
     - name: Tempo
       uid: tempo
       type: tempo
       access: proxy
       url: http://tempo:3200
   
     - name: Mimir
       uid: mimir
       type: prometheus
       access: proxy
       url: http://mimir:9009/prometheus
       jsonData:
         httpMethod: POST

   Mimir exposes a Prometheus-compatible query API, so Grafana uses the Prometheus data source type for it.

3. Mount the provisioning directory into the Grafana service.

   compose.yaml

   services:
     grafana:
       image: grafana/grafana:latest
       ports:
         - "3000:3000"
       volumes:
         - type: bind
           source: ./grafana/provisioning
           target: /etc/grafana/provisioning
           read_only: true

   Merge this service block with the existing stack definition instead of replacing the backend services.

4. Recreate the Grafana container so it reads the provisioning file.

   $ docker compose up -d grafana
   Container lgtm-grafana-1  Recreate
   Container lgtm-grafana-1  Started

5. List the data sources through the Grafana API.

   $ GRAFANA_URL=http://127.0.0.1:3000
   $ curl --silent \
     --user admin:admin \
     "$GRAFANA_URL/api/datasources"
   [
     {
       "id": 1,
       "uid": "loki",
       "name": "Loki",
       "type": "loki",
       "url": "http://loki:3100"
     },
     {
       "id": 2,
       "uid": "tempo",
       "name": "Tempo",
       "type": "tempo",
       "url": "http://tempo:3200"
     },
     {
       "id": 3,
       "uid": "mimir",
       "name": "Mimir",
       "type": "prometheus",
       "url": "http://mimir:9009/prometheus"
     }
   ]

   Replace admin / admin before saving this pattern in shared scripts or CI logs.

6. Open Connections → Data sources in Grafana.
7. Select each LGTM data source and click Save & test.
8. Open Explore and run one smoke query per backend.

   Use {job=~".+"} for Loki, a recent trace search for Tempo, and up or another known series for Mimir after metrics have been written.