The Gitlab chart comes with a handful of HorizontalPodAutoscaler resources by default, which make use of the `metrics.k8s.io` APIService to scale up and down deployments. Ensure the High Availability section of the Gitlab Architecture doc references the presence of these HorizontalPodAutoscalers and that every pod/deployment/statefulset in the Gitlab chart can be ran with more than one replica out of the box with the exception of gitaly. [Which has some (BigBang/P1) un-tested prerequisites to run as a cluster with High Availability.](https://docs.gitlab.com/ee/administration/gitaly/praefect.html) For Gitaly we have outlined in the docs/production.md document that bumping up the resources a fair amount is enough to cover for this lack of HA support.