Prometheus monitoring configuration with physical measurement resource

This document describes the role of the PhysicalMeasurement custom resource in the configuration of Chantico components. It explains the design and data model, particularly how it differs from other custom resources.

Whereas the MeasurementDevice custom resource describes a type of device and how it could be monitored (with SNMP authentication and walks), the PhysicalMeasurement custom resource describes the actual endpoint that is to be monitored, i.e., where the device is located. This is important to make use of the defined SNMP exporter authentications and modules to point at a specific IP address to actually obtain metrics. This is what Prometheus does to scrape the metrics: provide the parameters to the SNMP exporter to obtain the metrics from a specific device, with a selected authentication setting and module.

Fields

Note that the ip field may be anything that is resolvable by the SNMP exporter instance, including hostnames, fully qualified domain names and IP addresses. Hostnames are only used in mock/development/demonstration environments, where the SNMP exporter uses the Kubernetes domain name resolution to resolve the hostname to a service living in the same namespace, such as the mock SNMP device that is used for development and testing. The IP address needs to point to the monitored device on the subnet that the kubernetes cluster is able to reach, i.e., it requires that the subnet is attached and has routes defined on the cluster nodes. Often, such SNMP devices have their endpoints attached on a private management network, to avoid exposing them to the public internet and reducing IP address allocation for such purposes. In such cases, the management network needs to be stretched to the routers that the cluster nodes connect to, to allow them to access those devices. This is relevant during the deployment of chantico in production.