Tctl app ls output throws duplicates.

[nivasomu]

Expected behavior:
tctl app ls should output the same result as tsh app ls.

Here is the tsh app ls output

tsh apps ls | grep app
app                                            Discovered application in Kubernetes cluster ... TCP django       ClusterName=test-1,Environme... 

Current behavior:
When the discovery service discovers applications.

When tctl app ls command is executed. We see duplicate apps associated with all the agents deployed in other clusters.

Here is an example

tctl apps ls | grep app
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app..... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app..... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  
teleport-kube-agent-0 app                             TCP app.... tcp://app... ClusterName=test,Environment=Test... 16.4.3  

This brings confusion and challenges while registering the apps.

Bug details:

• Teleport version: 16.4.3

[zmb3]

I imagine this is intentional.

tctl is an administrative tool. Cluster administrators probably care to see how many agents are heartbeating for each app.

tsh is a user tool. Users just want to see what apps they have access to and don't care about the details.

[greedy52]

I imagine this is intentional.

tctl is an administrative tool. Cluster administrators probably care to see how many agents are heartbeating for each app.

tsh is a user tool. Users just want to see what apps they have access to and don't care about the details.

Yes. The first column is the host of each app_server heartbeat.

But why do they have the same hostname for the same app? Is it just a mockup text? If not, could you share tctl get app_server/app?