Knowledge Base

How to monitor Neo4j with Prometheus

Commencing with the release of Neo4j Enterprise 3.4, one can now use the open source monitoring tool Prometheus to monitor Neo4j. The following article details a basic Prometheus implementation to demonstrate the connectivity between Prometheus v2.2.1 and Neo4j.

After installing Prometheus, Neo4j needs to be configured via its neo4j.conf file with the following parameters:

# Enable the Prometheus endpoint. Default is 'false'.
metrics.prometheus.enabled=true
# The IP and port the endpoint will bind to in the format <hostname or IP address>:<port number>.
# The default is localhost:2004.
metrics.prometheus.endpoint=localhost:2004

and the prometheus.yml file should be configured as follows:

# A scrape configuration containing exactly one endpoint to scrape:
# Here it's Prometheus itself.
scrape_configs:
  # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config.
  - job_name: 'Neo4j-prometheus'

    # metrics_path: /metrics
    # scheme defaults to 'http'.

    static_configs:
      - targets: ['localhost:2004']

In the above configuration snippets, Prometheus has been installed on the same server as Neo4j. If you choose to install it on a different server, simply replace references to localhost above with the IP address of the Neo4j instance.

Launching the Prometheus browser at http://<IP of Prometheus Server>:9090 and clicking the menu option Status → Targets should display:

image

Returning back to the 3rd menu choice of Graphs one can define a graph to monitor a Neo4j metric. For example, in the detail below the graph represents the number of transactions started (neo4j_transaction_started):

image