Welcome to the Linux Foundation Forum!

Lab 3.4 - Deploy a Simple Application

rbraga
rbraga Posts: 10
edited January 10 in LFS258 Class Forum

I'm having trouble viewing the output of nginx via curl inside the master node, I'm using GCP and the firewall rule allows access to everything, what am I doing wrong?

when I use tcpdump it receives the request but there is no content from the webserver

Regards,
Rousseau

Comments

  • chrispokorni
    chrispokorni Posts: 1,161

    Hi @rbraga,

    Can you also provide screenshots of your Firewall rules and VPC settings?

    Regards,
    -Chris

  • rbraga
    rbraga Posts: 10

    Hi of course I'll show you


  • chrispokorni
    chrispokorni Posts: 1,161

    Your infra config looks ok.
    Is there an OS firewall that may be enabled?
    Are your nodes ready and all your control plane pods running ok?

    kubectl get nodes
    kubectl get pods -Ao wide

    Regards,
    -Chris

  • rbraga
    rbraga Posts: 10

    Hi Chris check out below

    I cleared the firewall rule

    I checked the firewall's rules, there were rules that ubuntu itself initializes, so I cleaned them all, but I still don't see the content using curl, I entered the pod shell using kubectl exec and I can see the output using localhost curl

    Regards,
    -Rousseau

  • serewicz
    serewicz Posts: 948

    Hello,

    As the pod appears to be running perhaps there is something else at issue. Use the Pod IP from the kubectl get pod -o wide and curl to test if you can see the web page via the PodIP.

    If that works, look at the endpoints, kubectl get ep and kubectl get svc. The service IP should forward traffic to the endpoint IP of your pod.

    How did you create the service? Did you use kubectl expose or via a YAML file. Delete the service, test that direct to Pod IP works, then create the service again using kubectl expose. Does it work then?

    Regards,

  • rbraga
    rbraga Posts: 10
    edited January 20

    Hi @serewicz
    I created the service as described in the lab's pdf, just using kubectl expose deployment / nginx, I also deleted the service and it doesn't work, do I have to change the default.conf of nginx ??

    I already did this pdf activity in GCP as in the cluster in my work the result is the same.

    I can't believe that I've been stuck in this activity for a few days trying to find out why the result doesn't come out the same as the PDF.

    regards

  • chrispokorni
    chrispokorni Posts: 1,161
    edited January 20

    Hi @rbraga,

    Your latest snapshots reveal that the issue may be caused by the way the networking is setup.

    Your Pod IPs are from the 10.234.x.x subnet. What is the size of the Pod network, as setup by your CNI plugin and as it was included in the kubeadm init command?

    Also, it seems that the Service Cluster IPs are also from the same network 10.234.x.x.
    If the Pod network and the Services network overlap, this may cause routing issues when the Service is attempting to forward the request to its Endpoint Pod(s).

    In a cluster the Node IP addresses, the Pod network from the CNI plugin, and the Service Cluster IP network should not overlap, they should all be distinct.

    Regards,
    -Chris

  • chrispokorni
    chrispokorni Posts: 1,161

    Hi @rbraga,

    On GCP, you seem to have the following config:

    • Your VPC is 10.21.0.0/16 so the expected VM IP addresses should be from this subnet (from the VPC Network details). The firewall rule to allow all traffic is also set for this VPC network.

    • Your Node IP addresses were 10.158.0.2 and 10.158.0.3 - not part of the VPC, and not under the firewall rule defined earlier.

    Was all this intentional? Or do you have different VPC networks, with different Firewall rules?

    Regards,
    -Chris

  • rbraga
    rbraga Posts: 10

    Hi @chrispokorni
    These last screenshots are from a my personal server that I built with rancher and it is not GCP, I'll do it at GCP again following the course video again to see if I can get the same result.

  • rbraga
    rbraga Posts: 10

    I deleted all the VM's and redid it all again and it worked.

    thanks for all the help

    Regards

Categories

Upcoming Training