Azure / AKS-Edge

Welcome to the Azure Kubernetes Service (AKS) Edge repo.
MIT License
53 stars 34 forks source link

AKS Edge Essentials single node deployment failure on connect to Arc[BUG] #164

Open clin4 opened 7 months ago

clin4 commented 7 months ago

Describe the bug I am following the steps from https://learn.microsoft.com/en-us/azure/aks/hybrid/aks-edge-quickstart to deploy a single node AKS Edge Essentials on my window 10 machine also from an Azure VM running windows 10 server. but failed to connect Arc.

To Reproduce Steps to reproduce the behavior:

  1. I follow this link to setup the mechine. [https://learn.microsoft.com/en-us/azure/aks/hybrid/aks-edge-howto-setup-machine]
  2. I follow this link to setup the single linux node k3s cluster [https://learn.microsoft.com/en-us/azure/aks/hybrid/aks-edge-howto-single-node-deployment]
  3. The cluster itself works, I can connect to it and deploy the azure vote app and test it use the NodePort on the vm
  4. I failed to follow this link to connect the k3s to azure arc [https://learn.microsoft.com/en-us/azure/aks/hybrid/aks-edge-howto-connect-to-arc] a. I created the resource group in my subscription b. I register all the required provider Microsoft.HybridCompute Microsoft.GuestConfiguration Microsoft.HybridConnectivity Microsoft.Kubernetes Microsoft.ExtendedLocation Microsoft.KubernetesConfiguration c. I have a service principle with the role of "Microsoft.Kubernetes connected cluster" d. I update the aksedge-config.json with all required value for the keys: ClusterName Location ResourceGroupName SubscriptionId TenantId ClientId ClientSecret e. I run the script [Connect-AksEdgeArc -JsonConfigFilePath .\aksedge-config.json], and get expected log-

[12/12/2023 11:06:49] AksEdge - Connecting cluster to Azure Arc

f. But I cannot get anything from the resource group

g. When I check the pods states in the azure-arc namespace, the kube-aad-proxy pod cannot start.. kube-aad-proxy-c4f67f59d-t7dv5 0/2 ContainerCreating 0 24m when I inspect the pod: I saw the following error: Warning FailedMount 2m32s (x19 over 25m) kubelet MountVolume.SetUp failed for volume "kube-aad-proxy-tls" : secret "kube-aad-proxy-certificate" not found

h. when I tried to list the secrets under the azure-arc namespaces, I cannot find it kubectl -n azure-arc get secret NAME TYPE DATA AGE azure-arc-connect-privatekey Opaque 1 27m

i. I uninstalled the arc component via this command and repeat the step e, the same issue again. az connectedk8s delete --name "cluster-name"--resource-group "my rg"

Expected behavior I am expecting to get the arc connect described from this link [https://learn.microsoft.com/en-us/azure/aks/hybrid/aks-edge-howto-connect-to-arc]

Screenshots If applicable, add screenshots to help explain your problem. 1 2

Environment (please complete the following information):

Try 1:

gshiva commented 6 months ago

@clin4 - did you increase the Linux Node size to have at least 4.5GB RAM and 20GB disk space?

I had a similar issue here #156

AKS-Edge team, can we make the scripts fail if the Azure Arc connection setup is attempted and the allocated memory is not at least 4.5GB?

As you can see, we don't read the doc very well :)

barnstee commented 6 months ago

I have the same issue and my single node cluster has 10GB of RAM and 40GB of storage...

barnstee commented 6 months ago

Switching back to Version 1.2.414.0 (K8s) resolves the error for me.