Starting with version 0.9.0 of Promregator, annotation-driven configuration of targets is supported if the underlying Cloud Foundry platform supports the CAPI V3 interface. Note that in these days the latter should not be a challenging requirement given the fact that the former interface, CAPI V2, has been deprecated.
Configuration follows the approach suggested by cf-for-k8s. See also that page fro examples how to set annotations.
These annotations are supported:
| Annotation | supported | Comments |
|---|---|---|
prometheus.io/scrape |
yes | must be set to "true" |
prometheus.io/path |
yes | overwrites any other path attribute in Promregator's main configuration, if provided |
prometheus.io/port |
no |
For the metadata to be evaluated, the corresponding target needs to have set the attribute kubernetesAnnotations to true (see also details at the configuration option page for this option). Note that if this option is set, any target which does not provide the necessary metadata annotations will not be selected for scraping (but ignored).
This provides the opportunity for an "opt-in" to scraping by applications without the need of modifying Promregator's configuration: If another application wants to get scraped, it may set prometheus.io/scrape to true (and provide a prometheus.io/path if applicable). On the next metadata update, the new application will be detected automatically and added to discovery.
An example Cloud Foundry application using these annotations would be deployed using a sample manifest like below:
---
applications:
- name: go-app-with-metrics
metadata:
annotations:
prometheus.io/scrape: "true"
prometheus.io/path: "/not/default/metrics"Given this configuration, the app would be scraped by promregator and the path
for scraping would be /not/default/metrics.