Picking up first running pod instead of first unknown status pod from list

[rvarunrathod]

It is sort of an edge case.
This will only happen when you have multiple instances of the same service.

1. When you have low resources, one pod is in a running state, and the other is in a pending state. It tries to port forward to the pending state pod.
   e.g.

NAME        READY   STATUS    
myapp-adfa   1/1     Running  
myapp-rasf   0/1     Pending    <- connection to this

2. I had a service with replications. When I was doing a rolling upgrade with new deployment, I had set wrong secret that pod could not find and got stuck in the ContainerCreating state. So there were older pods in a running state instead of connecting to them, it was trying to connect the stuck pod.
   e.g.

NAME        READY   STATUS    
myapp-adfa   1/1     Running  
myapp-rasf   0/1     ContainerCreating    <- connection to this

error upgrading connection: unable to upgrade connection: pod not found ("myapp_dev")
👓  Forwarder: lost port-forward connection trying to reconnect...
error upgrading connection: unable to upgrade connection: pod not found ("myapp_dev")
👓  Forwarder: lost port-forward connection trying to reconnect...

[eko]

Thank you for this fix @rvarunrathod.

I think a good approach for later would be to allow using a port-forward on a service (rather than on a pod directly). This will avoid selecting non-ready pods.

[rvarunrathod]

Hi @eko, Yes, port-forwarding to service is a good approach as it will make it fault-tolerant and increase availability, and we don't have to worry about which pod to pick. I will happily help you with implementation if you want me to.

[eko]

@rvarunrathod I do not have so much time for now so if you want to work on this, you're really welcome :)