For LFD259 2.3, What Am I Missing In the GCE Setup that is stopping curl from working?

mretfaster

Hello!
I'm stuck/failing at curling a pod during 2.3.

Here's what has succeeded -
I made the controlplane node & the worker node, on 2 different vms in GCE
I deployed a pod from the controlplane node and can see the pod via kubectl get pods.

Here's where I'm stuck -
I cannot curl the IP of the pod from neither the controlplane vm nor the worker vm.

I have 2 compute nodes:

I have a VPC setup, using the default setup which looks to work with the IP addresses in the region I chose for the VM instances:

I made a firewall rule that I believe allows all traffic to the nodes:

chrispokorni

Hi @mretfaster,

The introductory chapter of the course includes a video that guides the VM provisioning process on GCE, together with VPC and firewall creation tips to ensure the networking is properly configured.

The way the lab is set up, should allow you to curl both the pod IP and the service IP from either of the 2 nodes.

What is the output of kubectl get pod -A -o wide ?

Regards,
-Chris

mretfaster

Also for some more transparency of the k8s objects & commands used - note I've curled against the pod ip and the svc ip, both timeout. I assume only the ip of the svc should work
Here, I tried curling the svc ip:

mretfaster

Thanks for the note there -
I missed one of the setup details in the GCE video!! My VMs were not part of the VPC network.
After watching the video I was able to stop the cp + worker vms, edit their settings and add them to the network, and turn the vms back on. Then - curling the pod worked!!
Thanks again

andybeeching

Just here to say thanks for this info! Also on GCE I was running into an error where the VMs could not communicate with each other via the internal IPs. After much head-banging it turns out Cilium does funny things when using a 10.0.0.0 subnet.

Using 10.2.0.0 as in the GCE setup video fixed it.