This is an exporter that exposes information gathered from SNMP for use by the Prometheus monitoring system.

apt-get install libsnmp-python  # On older Debian-based distros.
apt-get install python-netsnmp  # On newer Debian-based distros.
yum install net-snmp-python     # On Red Hat-based distros.

pip install snmp_exporter

snmp_exporter [config_file] [port]

config_file contains what OIDs to scrape and how to process them. config_file defaults to snmp.yml. port defaults to 9116.

Visit http://localhost:9116/metrics?address=1.2.3.4 where 1.2.3.4 is the IP of the SNMP device to get metrics from. You can also specify a module parameter, to choose which module to use from the config file.

The snmp exporter needs to be passed the address as a parameter, this can be done with relabelling.

Example config:

scrape_configs:
  - job_name: 'snmp'
    target_groups:
      - targets:
        - 192.168.1.2  # SNMP device.
    relabel_configs:
      - source_labels: [__address__]
        regex: (.*?)(:80)?
        target_label: __param_address
        replacement: ${1}
      - source_labels: [__param_address]
        regex: (.*)
        target_label: instance
        replacement: ${1}
      - source_labels: []
        regex: .*
        target_label: __address__
        replacement: 127.0.0.1:9116  # SNMP exporter.

This setup allows Prometheus to provide scheduling and service discovery, as unlike all other exporters running an exporter on the machine from which we are getting the metrics from is not possible.

There are two components. An exporter that does the actual scraping, and a generator that creates the configuration for use by the exporter. Only the exporter is written so far.

This is to allow for customisation of what's done during the scrape as many special cases are expected. The varying levels of SNMP MIB-parsing support across different langauges also means that a single language may not be practical.