stack-orchestrator/stack_orchestrator/data/stacks/monitoring

monitoring

• Instructions to setup and run a Prometheus server and Grafana dashboard
• Comes with the following built-in exporters / dashboards:
  • Chain Head Exporter - for tracking chain heads given external ETH RPC endpoints
  • Watchers dashboard
  • Prometheus Blackbox - for tracking HTTP endpoints
  • NodeJS Application Dashboard - for default NodeJS metrics
  • PostgreSQL Database - for monitoring Postgres dbs
  • Node Exporter Full - for monitoring system metrics
• See monitoring-watchers.md for an example usage of the stack with pre-configured dashboards for watchers

Setup

Clone required repositories:

laconic-so --stack monitoring setup-repositories --git-ssh --pull

Build the container images:

laconic-so --stack monitoring build-containers

Create a deployment

First, create a spec file for the deployment, which will map the stack's ports and volumes to the host:

laconic-so --stack monitoring deploy init --output monitoring-spec.yml

Ports

Edit network in spec file to map container ports to same ports in host:

...
network:
  ports:
    prometheus:
      - '9090:9090'
    grafana:
      - '3000:3000'
...

Data volumes

Container data volumes are bind-mounted to specified paths in the host filesystem. The default setup (generated by laconic-so deploy init) places the volumes in the ./data subdirectory of the deployment directory. The default mappings can be customized by editing the "spec" file generated by laconic-so deploy init.

---

Once you've made any needed changes to the spec file, create a deployment from it:

laconic-so --stack monitoring deploy create --spec-file monitoring-spec.yml --deployment-dir monitoring-deployment

Configure

Prometheus Config

• Add desired scrape configs to prometheus config file (monitoring-deployment/config/monitoring/prometheus/prometheus.yml) in the deployment folder; for example:

  ...
  - job_name: <JOB_NAME>
    metrics_path: /metrics/path
    scheme: http
    static_configs:
      - targets: ['<METRICS_ENDPOINT_HOST>:<METRICS_ENDPOINT_PORT>']
  

• Node exporter: update the node job to add any node-exporter targets to be monitored:

  ...
  - job_name: 'node'
    ...
    static_configs:
      # Add node-exporter targets to be monitored below
      - targets: [example-host:9100]
        labels:
          instance: 'my-host'
  

• Blackbox (in-stack exporter): update the blackbox job to add any endpoints to be monitored on the Blackbox dashboard:

  ...
  - job_name: 'blackbox'
    ...
    static_configs:
      # Add URLs to be monitored below
      - targets:
        - <HTTP_ENDPOINT_1>
        - <HTTP_ENDPOINT_2>
  

• Postgres (in-stack exporter):

  • Update the postgres job to add Postgres db targets to be monitored:

    ...
    - job_name: 'postgres'
      ...
      static_configs:
        # Add DB targets below
        - targets: [example-server:5432]
          labels:
            instance: 'example-db'
    

  • Add database credentials to be used in auth_modules in the postgres-exporter config file (monitoring-deployment/config/monitoring/postgres-exporter.yml)

Note: Use host.docker.internal as host to access ports on the host machine

Grafana Config

Place the dashboard json files in grafana dashboards config directory (monitoring-deployment/config/monitoring/grafana/dashboards) in the deployment folder

Env

Set the following env variables in the deployment env config file (monitoring-deployment/config.env):

# For chain-head exporter

# External ETH RPC endpoint (ethereum)
# (Optional, default: https://mainnet.infura.io/v3)
CERC_ETH_RPC_ENDPOINT=

# Infura key to be used
# (Optional, used with ETH_RPC_ENDPOINT if provided)
CERC_INFURA_KEY=

# External ETH RPC endpoint (filecoin)
# (Optional, default: https://api.node.glif.io/rpc/v1)
CERC_FIL_RPC_ENDPOINT=

# Grafana server host URL (used in various links in alerts, etc.)
# (Optional, default: http://localhost:3000)
GF_SERVER_ROOT_URL=

Start the stack

Start the deployment:

laconic-so deployment --dir monitoring-deployment start

• List and check the health status of all the containers using docker ps and wait for them to be healthy

• Grafana should now be visible at http://localhost:3000 with configured dashboards

Clean up

To stop monitoring services running in the background, while preserving data:

# Only stop the docker containers
laconic-so deployment --dir monitoring-deployment stop

# Run 'start' to restart the deployment

To stop monitoring services and also delete data:

# Stop the docker containers
laconic-so deployment --dir monitoring-deployment stop --delete-volumes

# Remove deployment directory (deployment will have to be recreated for a re-run)
rm -rf monitoring-deployment