[teleport] kops fails due to inconsistent cluster names

[Nuru]

what

When using Teleport tsh login to connect to a cluster, although kubectl works fine, kops does not, because, in general, the Teleport name of the cluster is not the same as the kops cluster name. This leads to an error like

 ⧉  staging (staging:default)+
 √ : (cpco-staging-admin) helmfiles ⨠ kops validate cluster
Validating cluster us-west-2.staging.cpco.io

Cannot load kubecfg settings for "us-west-2.staging.cpco.io": context "us-west-2.staging.cpco.io" does not exist

why

kops needs the name of the cluster in 2 places: its configuration in its state store and the context setting in KUBECONFIG. It has a --name flag to set the name, but it still expects it to be the same in both places. Typically, however, the Teleport name of the cluster is different due to Teleport requirements.

workaround

There is a workaround. After tsh login you can run:

kubectl config rename-context $(kubectl config current-context) $KOPS_CLUSTER_NAME

This enables kops to find the correct context and does not bother kubectl because technically the context name is independent of the cluster name.