Monitoring the model-server

Besides the afore mentioned use cases, the samples also include a model-server monitoring setup. This setup uses the /metrics endpoint provided by the model-server by default.

Instructions

Start the components

Command to start the monitoring profile
docker-compose --profile monitoring up

This command starts the following components:

If you use a locally running `model-server, make sure you picked the correct instance at the top of the Grafana dashboard. Otherwise, the dashboard might show no data.

Depending on your operating system or docker setup, you might need to adjust the targets defined in the prometheus.yml and datasource.yml.

Also, the docker network setup might need adjustments in the docker-compose.yaml, e.g. when ports are already occupied or if you run a model-server locally and not via docker.

Engage with the model-server

In order to see noteworthy changes in the monitoring setup, you can simply direct any request towards the model-server. This process can be simplified by using the Swagger UI included in the model-server over at http://127.0.0.1:28101/swagger.

You can also start the other use cases in parallel.

If you want to start everything at the same time, you can simply start the all profile via

Command to start the monitoring profile
docker-compose --profile all up