Alerting on metrics


Currently doing some initial research into elastic for use among other things as an infrastructure monitoring tool. Metricbeat + elastic appear to give us exactly the sort of thing we want, so I started the 14 day trial to test out alerting.

How should watcher be used for infrastructure monitoring? How do I apply a rule to a group of servers e.g. when CPU above 80% for 5 minutes for this group of endpoints, alert. When CPU at 100% for 5 minutes for this group of endpoints, alert.

From what I can tell, the only way that may be possible to achieve this is to do it in straight up JSON, which is disappointing if it is not possible in the UI.


@stuart475898 We are currently working on the integrated alerting feature for the Metric UI. It’s being built using the new Kibana alerting platform which expected to ship soon.

Here is our meta issue for tracking the work:

Watcher is an option but I would probably recommend waiting if that’s possible.