NetKAN-Status-Exporter

NetKAN Status Exporter

A Prometheus exporter for the NetKAN infrastructure status of the KSP-CKAN project.

License: GPLv3

Usage

The exporter listens on :9811 by default, it can be overwritten using the --web.listen-address commandline option, or NETKAN_STATUS_EXPORTER_WEB_LISTEN_ADDRESS environment variable. Add a Prometheus scrape config to scrape it like this:

scrape_configs:
  - job_name: 'netkan-status'
    scrape_interval: 30m
    scrape_timeout:  20s
    static_configs:
      - targets: [ 'exporter.host:9811' ]

The NetKAN inflator basically only runs every 30 minutes, so it makes sense to set the scrape_interval just as low to reduce load on the CKAN infrastructure. However, Prometheus considers metrics not updated for more than 5 minutes stale and won't return any value for them, you can work around this by using max_over_time(netkan_some_metric[30m]). You need a relatively high timeout (e.g. 20s), since downloading and parsing the JSON can take quite a long time.

Docker

There's a Docker image available at registry.gitlab.com/dasskelett/netkan-status-exporter and dasskelett/netkan-status-exporter. It's built using golang:alpine, and the final image is based on alpine:3.13 and thus only a few MiB in size.

Third-party packages

This project uses

• Prometheus Go client library (Guide), Apache-2.0
• Kingpin command line and flag parser, MIT
• goccy/go-json, MIT for optional faster JSON decoding. Warning: Experimental, has memory leaks