The /metrics
endpoint should be enabled by default (couldn't find any documentation on enabling/disabling this feature, seems to be always on)
Prometheus needs to be configured to scrape the metrics endpoint from itself:
scrape_configs:
# The job name is added as a label `job=<job_name>` to any timeseries scraped from this config.
- job_name: 'prometheus'
# Override the global default and scrape targets from this job every 5 seconds.
scrape_interval: 5s
static_configs:
- targets: ['localhost:9090']
Taken from the getting started guide.
I guess the most relevant metric from the Prometheus endpoint regarding disk space usage would be: prometheus_tsdb_storage_blocks_bytes
But for fine tuning and profiling you might want to install prometheus node exporter as well.