Azure/helm-charts

Prerequisite README.md incorrectly states that AKS will not work

Closed this issue · 2 comments

The prerequisites README.md file incorrectly states that AKS doesn't support aggregated APIs and won't support service-catalog installation. Now that AKS does support this, we should update the prerequisites file.

Does it work? I tried following the steps and the Service Catalog Controller Manager pod never comes up.

kubectl get pods --namespace catalog --watch

NAME                                                  READY     STATUS    RESTARTS   AGE

catalog-catalog-apiserver-74cff45684-mtp28            2/2       Running   0          31s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Running   0          31s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Running   1         37s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Error     1         38s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Running   2         48s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Error     2         49s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       CrashLoopBackOff   2         50s

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Running   3         1m

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Error     3         1m

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       CrashLoopBackOff   3         1m

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Running   4         1m

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       Error     4         1m

catalog-catalog-controller-manager-8458c655cf-47gqp   0/1       CrashLoopBackOff   4         1m

It worked on a newly created AKS cluster, not my existing cluster (v 1.8.1)

Hey @sabbour, it should be working on AKS. As you seem to have discovered, however, the change to support api server aggregation was not retroactively applied to existing AKS clusters that existed prior to the AKS team fixing that issue. This is why your existing cluster failed, while the newly created cluster was successful.

I'll try to determine when that fix was made and include the date in the docs.