CPU saturation forecasting on cpu, shard_cpu, single_node_cpu
We just saw an example where we get 3 similar looking forecasts:
-
cpu
: https://gitlab.com/gitlab-com/gl-infra/capacity-planning-trackers/gitlab-com/-/issues/1757+ -
shard_cpu
: https://gitlab.com/gitlab-com/gl-infra/capacity-planning-trackers/gitlab-com/-/issues/1758+ -
single_node_cpu
: https://gitlab.com/gitlab-com/gl-infra/capacity-planning-trackers/gitlab-com/-/issues/1759+
This sounds related to what &1287 is aiming to solve. We may be able to review these saturation points and decide if we retain them or replace with an automatic dimensional forecast.