July 2019
Intermediate to advanced
502 pages
14h
English
You can raise alerts by configuring rules in the Prometheus server configuration file. Those alerts are handled by the alert manager which decides, based on its configuration, what to do about them. Here is an example:
groups:- name: link-manager rules: - alert: SlowAddLink expr: link_addlink_summary{quantile="0.5"} > 5 for: 1m labels: severity: critical annotations: description: the AddLink() method takes more than 5 seconds for more than half of the request in the last minute summary: the AddLink() method takes too long
The rule has an expression, which, if true, triggers the alert. There is a period of time (1 minute here) where the condition must be true, so that you can avoid triggering one-off anomalies ...