kubectl periodically fails to connect to the API server with no such host

[alika]

What happened:
I periodically cannot connect to our AKS cluster with kubectl:

$ kubectl cluster-info

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.
Unable to connect to the server: dial tcp: lookup global-260e2f2d.hcp.westus2.azmk8s.io: no such host

$ kubectl get no
Unable to connect to the server: dial tcp: lookup global-260e2f2d.hcp.westus2.azmk8s.io: no such host

After about 5-10 minutes it successfully connects. This is occurring several times a day, and being experienced by multiple engineers.

What you expected to happen:
I expect to be able to connect to the AKS consistently with kubectl.

How to reproduce it (as minimally and precisely as possible):
kubectl cluster-info

Anything else we need to know?:

• This started to occur after I recreated the AKS cluster with the same name (delete, create).
• I'm using the Terraform for AKS (azurerm_kubernetes_cluster resource) to create and maintain the AKS cluster.
• There are similar no such host errors in the tunnelfront pod logs
• Similar to kubectl stopped working with ***eastus.azmk8s.io: no such host #104

Environment:

• Kubernetes: 1.11.8
• Size: 5 x Standard_DS3_v2
• Workloads: jenkins, cert-manager, harbor, cockroachdb

[jnoller]

@alika please open an azure support ticket in the portal so the team can RCA the issue, also I'd recommend deleting the subscription ID/region from your ticket ASAP (this is why we ask for support requests so this information doesn't get out).

[alika]

@jnoller, unfortunately this particular subscription does not have a support plan in which i can create technical support requests.

[robinkb]

FYI, GitHub saves a history of all edits made. That means your information is still easily retrieved.

[alika]

@robinkb, do you have a suggestion on how to purge this somehow? Or do I need to scrap this cluster?

[robinkb]

@alika Please refer to the GitHub documentation.

The problem is not really the ID of your cluster, but your subscription ID. You cannot change that by scrapping the cluster. See this StackOverflow question on why it might be considered sensitive.

There is no way to change the ID of a subscription, I don't think. So it is up to you to judge whether or not the ID of this particular subscription must be kept secret. If it is, I think that you will have to delete the subscription and create a new one.

[alika]

@robinkb, Thanks for the delete history reference.

[jnoller]

Duplicate of #232

[speciesunknown]

@jnoller

This is not a duplicate of #232

The error here is "no such host" whereas #232 is about timeouts.

[sappojisetty]

I'm trying to create a kubernetes cluster and the namespace using terraform and I see this issue. Can somebody help me to find solution for this?
I'm using Terraform v0.12.21

[m-osh]

I just encountered the same issue in v0.12.21 as well
@sappojisetty did you figure it out?

[sappojisetty]

I just encountered the same issue in v0.12.21 as well
@sappojisetty did you figure it out?

I fixed this issue by running the command kubectl config use-context <cluster-name> before applying the terraform.