Tweak ES disk space saturation points (!2077) · Merge requests · GitLab.com / Runbooks

Craig Furman requested to merge craigf/es-single-node-disk-space into master Apr 03, 2020

elastic_disk_space was actually a worst-case node metric, like single_node_cpu. It's fixed to actually represent the average disk space used across nodes.

Re-add elastic_single_node_disk_space as its own saturation point.

The dashboard saturation detail prom query is identical to the non-single-node one, because currently all saturation detail charts are not aggregated across nodes (unlike the component-level recording rules). Even though displaying 2 identical charts may seem odd, it is currently required, and we plan to add saturation detail chart links to our saturation alerting. Finding out which node is saturated will be the first thing a responder will want to do anyway, so aggregation across nodes is not desirable here.

Tweak ES disk space saturation points

Merge request reports