ingress-nginx on design.gitlab.com and version.gitlab.com are a SPOF
Corrective action for production#5259 (closed)
If we look at the deployment model for ingress-nginx in the version.gitlab.com and design.gitlab.com environment, we can see the ingress controller is a single pod
$ kubectl -n gitlab-managed-apps get deployment/ingress-nginx-ingress-controller
NAME READY UP-TO-DATE AVAILABLE AGE
ingress-nginx-ingress-controller 1/1 1 1 639d
$ kubectl -n gitlab-managed-apps get hpa
No resources found.
This means that during node operations (or unexpected node failures), if the single pod goes down, these applications will go offline and be unavailable.
We should change the deployment of these so that there are a minimum of two pods for redundancy.
As they are managed by Gitlab Managed Apps which is deprecated, I think manually editing the objects to have a replica of 2 is likely sufficent.
/cc @devin