Apollo Demo Server

This repo is built as a sandbox example repository for demonstrating features and capabilities of the Apollo eco-system, primarily focussed on the capabilities of the Apollo Router. Below is an overview of the project architecture and each of the different services with it's associated port.

Start the Subgraphs

There are currently 2 different example subgraphs built out with dummy responses (User and Demo) these can be started by using the following commands:

cd subgraph-demo
npm start

cd subgraph-user
npm start

Compose supergraph locally

This repo is built to run standalone and therefore is built on the assumption that Supergraph composition is happening locally. This can be achieved by running the following in the root directory. Note that the output file and config file reference can be updated in the underlying command defined in the package.json.

npm run compose

Start the router locally

This project is built to leverage a locally running binary instance of the Router. This can be installed and then run using the following commands in the root directory:

# From project root

# Will install the latest available version for Mac, the underlying request can be modified in package.json
npm run router:install

# This will run the installed local binary on the assumption a locally composed supergraph has been created
npm run router:local

Run Prometheus

This repo contains a working example of collecting metrics with Prometheus. There are some nuances to how prometheus collects metrics and how Router makes these metrics available. Router doesn't connect to Prometheus directly, instead it exposes the appropriately formatted metrics for prometheus on a specific port. Below is the config for making this possible within the router-config.yaml file in the root of the repo:

telemetry:
  metrics:
    prometheus:
      enabled: true
      listen: 0.0.0.0:9095
      path: /metrics

This config makes appropriately formatted metrics available on port 9095 from the router. It's possible to test this by simply running the router and navigating to http://localhost:9095/metrics. The next step is to have a running instance of Prometheus with this port configured as a target to scrape the available metrics. Once you have a running instance of Prometheus after following the Prometheus Getting Started steps, it is possible to configure the instance of Prometheus programatically by adding the following to the prometheus.yml config file used on startup:

scrape_configs:
    static_configs:
      - targets: ['0.0.0.0:9090', '0.0.0.0:9095']

Once this is in place, you should be able to navigate to the Prometheus GUI (i.e. http://localhost:9090/graph) and see the metrics from Router being pulled through.

Grafana Setup

Once you have Prometheus running and collecting metrics from Router successfully, Grafana is easily configured as it has native support for Prometheus. The key step for integrating with Prometheus is adding it as a data source as shown below:

Note: Be careful to target the running instance of Prometheus (Port: 9090) and not the exposed metrics from Router directly (Port: 9095).

Once this has been configured you should be able to build custom dashboards as required by leveraging the assigned data source.

Start the Open Telemetry Collector

This repo also has an example of how to collect Open Telemetry data from Router. the first step is to install a local instance of the Open Telemetry Collector by using the following commands:

# Download Open telemetry Collector for Mac M1
curl -O -L https://github.com/open-telemetry/opentelemetry-collector-releases/releases/download/v0.63.0/otelcol_0.63.0_darwin_arm64.tar.gz
tar -xvf otelcol_0.63.0_darwin_arm64.tar.gz

# Run collector with the local config
npm run otlp:start

The Open Telemetry Collector is configured via the otlp-config.yaml file in the root of the directory. Open Telemetry Collector has a huge array of available integrations and capabilities, for full configuration details visit the Open Telemetry Documentation.

In order to check the collector has been configured correctly you can query the output using the following endpoint:

http://localhost:8888/metrics

Start the gRPC Server to consume output of Open Telemetry Collector

One use case demonstrated in this repo is running a custom service that exposes a gRPC endpoint for receiving Open Telemetry Data from the Open Telemetry Collector. This service currently takes the output that is periodically sent from the Open Telemetry Collector and writes it to a file locally with the format TIMESTAMP-telemetry.json. You can run this service by running the following commands:

cd grpc-receiver
npm start

Note: this service will not produce output unless the Open Telemetry Collector is also running and configured as required.