Hands-On Infrastructure Monitoring with Prometheus

Federation configuration

A running Prometheus server exposes a special endpoint served at /federate. This endpoint allows for the retrieval of time series that match one or more instant vector selectors (which we covered in Chapter 7, Prometheus Query Language – PromQL). To make these mechanisms clear, we provided a very simple example in our test environment. Each one of the shard instances has a recording rule producing aggregate metrics representing the number of HTTP requests to exporters, illustrated in the following code block:

vagrant@shard01:~$ cat /etc/prometheus/rules.ymlgroups:  - name: recording_rules    rules:      - record: job:promhttp_metric_handler_requests:sum_rate1m expr: sum by (job) (rate(promhttp_metric_handler_requests_total[1m])) ...