Welcome to the home of `prometheus-ganeti-exporter. You can use this software to publish Ganeti cluster statistics to Prometheus. It has been initially developed by Wikimedia Foundation and is now part of the Ganeti project.
You can use the provided systemd service file to run the service in the background. Please create or use a non-privileged system user for the service to avoid running it as root! The RAPI user requires read-only permissions for the exporter to work.
The service expects its configuration file in /etc/ganeti/prometheus.ini (an alternative path can provided through the --config parameter). Please see prometheus.ini.example for an example configuration. Please make sure the file is only readable to the service user as it contains RAPI credentials.
You can configure the exporter to include data from Ganeti htools (currently hbal and hspace). The default setting is to not query these tools, as they need to be available on the machine the exporter is running, and they also might take a long time (especially hspace tends block for a while when running on unbalanced clusters). The exporter will kill these processes when they reach the configured refresh interval and log this as an error.
Please beware that htools will connect to the cluster using RAPI. Unfortunately this will reveal your RAPI credentials in the process list as they need to passed on the command line. If you plan to run the exporter from a node which is not itself a Ganeti node, you can either build htools from the Ganeti source or install the package ganeti-htools if you are on Debian. The latter will only install htools without Ganeti itself.
Metrics obtained from hbal include the current cluster score as well as the achievable cluster score. This allows you to monitor for clusters which are in need of rebalancing. You may specify additional parameters to hbal as e.g. --exclusion-tags using the exporter configuration file.
hspace simulates the creation of instances with a given disk template and a given size (specified as $storage,$memory,$vcpus). Currently, the only metric exported is the number of instances that could be allocated with the given parameters.
You may set the logging level using the argument --loglevel [error|warning|info|debug]. Please note that debug logging might leak sensitive data like your RAPI credentials. Use this setting with caution! The default loglevel is set to warning.
If you use Prometheus with Grafana you may want to take a look at the example overview dashboard or the example details dashboard. All dashboards in this repository are released under the BSD 2-Clause License.