Welcome to the Linux Foundation Forum!

Lab 9.2

reactivan
reactivan Posts: 5
edited June 2022 in LFS258 Class Forum

I have problems with this Lab. I work on a local setup with one CP and one Worker Node. Lab 9.1 is successfully installed. Now I can't go any further.

[[email protected] ~]$ less /etc/hosts
192.168.99.200 k8s-master-01.example.org k8s-master-01
192.168.99.201 k8s-worker-01.example.org k8s-worker-01

192.168.99.200 k8s-endpoint.example.org  k8s-endpoint
[[email protected] ~]$ kubectl -n accounting describe services
Name:                     service-lab
Namespace:                accounting
Labels:                   system=secondary
Annotations:              <none>
Selector:                 system=secondary
Type:                     NodePort
IP Family Policy:         SingleStack
IP Families:              IPv4
IP:                       10.101.211.61
IPs:                      10.101.211.61
Port:                     <unset>  8080/TCP
TargetPort:               8080/TCP
NodePort:                 <unset>  30977/TCP
Endpoints:                192.168.243.180:8080,192.168.243.189:8080
Session Affinity:         None
External Traffic Policy:  Cluster
Events:                   <none
[[email protected] ~]$kubectl cluster-info
Kubernetes control plane is running at https://k8s-endpoint.example.org:6443
CoreDNS is running at https://k8s-endpoint.example.org:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy
[email protected] ~]$ curl https://k8s-endpoint.example.org:30977
curl: (7) Failed to connect to k8s-endpoint.example.org port 30977: Connection refused
[email protected] ~]$ curl 127.0.0.1:30977
curl: (7) Failed to connect to 127.0.0.1 port 30977: Connection refused

It is the same when i am on the worker node. I have no open port 30977 on both nodes. I checked that with 'netstat'

So i have done the same on an eks to check this in a fast way. The result is the same.

Something went wrong here, do i need a load-balancer? Or is there something else missing that the learners should find out for themselves? In other labs I had the impression a few times that you were installing something like that, I can't find the problem here.

Best Answer

  • oleksazhel
    oleksazhel Posts: 57
    Answer ✓

    @reactivan You simply didn't change port from 8080 to 80 -- that's why you don't have access via NodePort (nginx itself doesn't listen on 8080). See lab 9.1 p13.

    You can't see that port is listening via netstat, because it is routed via iptables.

Answers

  • leopastorsdg
    leopastorsdg Posts: 14
    edited June 2022

    Hello @reactivan

    On Lab Exercise 9.2, it is expected that we use curl with HTTP protocol, not HTTPS. On the other hand, it is important to use "http://" on your command (not only IP address).

    Have you tried "curl http://:NodePort"?

    Leo

  • reactivan
    reactivan Posts: 5
    edited June 2022

    At fiist, sorry for the double post, i had problems with the markdown for the code examples. When I left the forum I thought it was only one post.

    I have not tried your command curl http://:30977 -but i tried it now for fun, it doesn't work. Also these::

    curl http://k8s-endpoint.example.org:30977
    curl http://k8s-endpoint:30977
    curl http://k8s-master-01:30977
    curl http://k8s-worker-01:30977
    curl http://k8s-master-01.example.org:30977
    curl http://k8s-worker-01.example.org:30977
    curl http://127.0.0.1:30977
    

    curl: (7) Failed to connect to * 30977: Connection refused
    It doesnt't work for me.

    As i say before, i checked with netstat on both nodes:

    # netstat -tulpn
    Proto Recv-Q Send-Q Local Address           Foreign Address         State       PID/Program name
    tcp        0      0 127.0.0.1:10257         0.0.0.0:*               LISTEN      1693/kube-controlle
    tcp        0      0 0.0.0.0:179             0.0.0.0:*               LISTEN      2541/bird
    tcp        0      0 127.0.0.1:10259         0.0.0.0:*               LISTEN      1713/kube-scheduler
    tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN      757/sshd
    tcp        0      0 127.0.0.1:41383         0.0.0.0:*               LISTEN      761/containerd
    tcp        0      0 127.0.0.1:10248         0.0.0.0:*               LISTEN      981/kubelet
    ...
    

    there is no open Port 30977

  • reactivan
    reactivan Posts: 5

    It works now with a simpler setup:

    kubectl run echoserver --image=k8s.gcr.io/echoserver:1.10 --restart=Never --port=8080 -l app=echoserver
    kubectl create service nodeport echoserver --tcp=5005:8080
    

    This works with all commands like:

    curl http://k8s-endpoint.example.org:31861
    curl k8s-endpoint:31861
    curl 192.168.99.200:31861
    

    But i have to understand that i can't see the nodeport 31861 with netstat.

  • reactivan
    reactivan Posts: 5

    @oleksazhel said:
    @reactivan You simply didn't change port from 8080 to 80 -- that's why you don't have access via NodePort (nginx itself doesn't listen on 8080). See lab 9.1 p13.

    You can't see that port is listening via netstat, because it is routed via iptables.

    now I started again with 8.1 and understood the whole thing. Many Thanks!

Categories

Upcoming Training