Welcome to the Linux Foundation Forum!

Lab 7.2 Ingress Controller - Service available via internal ip, not via ingress

tknoops
tknoops Posts: 7
in LFD259 Class Forum

Hi,

I'm running into an issue with Lab 7.2. I've managed to create the ClusterRole and ClusterRole binding as well as the traefik ingress controller. After doing this the ingress controller is up but on the external IP address I get a Connection Refused message from curl.

ip address for this vbox host:

2: enp0s3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
link/ether 08:00:27:99:32:3f brd ff:ff:ff:ff:ff:ff
inet 10.10.10.121/24 brd 10.10.10.255 scope global dynamic enp0s3
valid_lft 54906sec preferred_lft 54906sec
inet6 fe80::a00:27ff:fe99:323f/64 scope link
valid_lft forever preferred_lft forever

failing curl command:

[email protected]:~/LFD259/labs/7/app2$ curl -H "Host: www.example.com" http://10.10.10.121
curl: (7) Failed to connect to 10.10.10.121 port 80: Connection refused

Ingress exists, but it has no address. I suspect this might be the problem as the describe shows that the default-http-backend was not found (but I have no idea how to solve that...).

[email protected]:~/LFD259/labs/7/app2$ kubectl get ingress
NAME CLASS HOSTS ADDRESS PORTS AGE
ingress-test www.example.com 80 21h

[email protected]:~/LFD259/labs/7/app2$ kubectl describe ing ingress-test
Name: ingress-test
Namespace: default
Address:
Default backend: default-http-backend:80 ()
Rules:
Host Path Backends
---- ---- --------
www.example.com
/ secondapp:80 (192.168.148.24:80)
Annotations: kubernetes.io/ingress.class: traefik
Events:

I'm using VirtualBox so as mentioned in the labs, there maybe issues when using LoadBalancer instead of NodePort. The service is available on NodePort and running correctly:

[email protected]:~/LFD259/labs/7/app2$ kubectl get svc
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kubernetes ClusterIP 10.96.0.1 443/TCP 20d
nginx ClusterIP 10.105.71.234 443/TCP 19d
registry ClusterIP 10.105.67.82 5000/TCP 19d
secondapp LoadBalancer 10.110.213.97 80:32000/TCP 21h

[email protected]:~/LFD259/labs/7/app2$ kubectl describe svc secondapp
Name: secondapp
Namespace: default
Labels: run=my-nginx
Annotations:
Selector: example=second
Type: LoadBalancer
IP: 10.110.213.97
Port: 80/TCP
TargetPort: 80/TCP
NodePort: 32000/TCP
Endpoints: 192.168.148.24:80
Session Affinity: None
External Traffic Policy: Cluster
Events:

[email protected]:~/LFD259/labs/7/app2$ curl http://10.110.213.97
<!DOCTYPE html>


Welcome to nginx!

body { width: 35em; margin: 0 auto; font-family: Tahoma, Verdana, Arial, sans-serif; }


Welcome to nginx!

If you see this page, the nginx web server is successfully installed and working. Further configuration is required.

For online documentation and support please refer to nginx.org.
Commercial support is available at nginx.com.

Thank you for using nginx.


Any ideas what I'm missing here?

Comments

  • chrispokorni
    chrispokorni Posts: 1,842

    Hi @tknoops,

    You mentioned this:

    I'm using VirtualBox so as mentioned in the labs, there maybe issues when using LoadBalancer instead of NodePort

    Yet your service is still a LoadBalancer type. Did you try changing it to a NodePort type instead?

    Regards,
    -Chris

  • tknoops
    tknoops Posts: 7

    HI @chrispokorni , completely forgot to mention that but yeah I did try that but to no effect. The labs don't say I have to change it back though as a NodePort will always be available even though there is no LoadBalancer available (like in non cloud setups like mine).

  • tknoops
    tknoops Posts: 7

    I removed everything, tried to switch to Traefik image 2.2.1 which got the traefik pods going into a error crash backoff loop. Switched back to image 1.17.3 (without removing anything but the serviceaccount/daemonset/service) and it worked. Not sure how as I've redeployed everything a few times already, but hope this helps someone in the future.

  • chrispokorni
    chrispokorni Posts: 1,842

    The lab does not use traefik v2. If you read carefully step 4 of the ingress exercise, it states that v1.7.13 is used instead.

  • tknoops
    tknoops Posts: 7

    I'm aware of that but since 1.7.13 wasn't working either I was wondering if 2.2.1 did work. It didn't, but it did make me redeploy everything which somehow solved the issue :smile:

    For future references, address is still empty while I have both backend services available via the Traefik ingress service, so that wasn't part of the problem.

  • serewicz
    serewicz Posts: 1,000

    Hello,

    Perhaps if we troubleshoot from the pod out we can figure out where the issue may be.

    If you curl to the pod ip it works, 192.168.xx.yy? I think yes.
    If you curl to the endpoint of the service secondapp it works, 10.110.213.97? I think yes.

    So we know the web server is running and the service is working. On to if the Ingress controller is working. Do you see any messages for the Ingress controller pods or daemonset? If you run kubectl describe against each object are there any warnings or other messages that could help us to the next step?

    Your previous output indicates the Ingress pods are not getting IP addresses. As the IP is given via the API server that makes me wonder if there is an overall networking issue - not a Kubernetes issue.

    So, did you open up every port on every interface? VirtualBox locks down interfaces a bit and you may need to turn each one into promiscuous mode.

    From the host if you try to nc a VMs primary IP at port 80, does it connect to anything?

    Use tcpdump or wireshark to see where the traffic is going, both from the host as well as from the VM perspective?

    Troubleshooting network issues with VirtualBox can be tricky because there are several possible ways you set up the VB network. As well the host may have some firewall rules in place that are blocking the traffic. If nothing else works at least we can rule in or out that aspect with tcpdump information.

    Regards,

  • mmckenzie
    mmckenzie Posts: 3

    Hello, I think I can help here in case anyone else runs into this.

    The lab instructs you to run the following : [email protected]: ̃$ ip a

    The command will show you the ip address for the master node. What you really want is the ip address for the worker node since that is where your ingress controller is running.

    The documentation should instruct you to open a new terminal on the worker node and perform: [email protected]worker: ̃$ ip a

    Once I had the worker ip the request to host example.com worked.

  • serewicz
    serewicz Posts: 1,000

    Hello,

    This is incorrect. The ingress pod is running on all nodes. If it is not for your system perhaps you missed the step which removes the master taint, or other issue.

  • nkerr
    nkerr Posts: 10

    Hello,
    I seem to be having a similar issue, or perhaps a perceived issue.
    Essentially the command ip a does not present a usable IP for the curl command. After setting up ingress, ingress controller and such I end up with the ingress-test as follows

    Name:             ingress-test
Namespace:        anthill
Address:          
Default backend:  default-http-backend:80 (<error: endpoints "default-http-backend" not found>)
Rules:
Host             Path  Backends
 ----             ----  --------
www.example.com  
                  /   secondapp:80   192.168.77.199:80)
Annotations:       kubernetes.io/ingress.class: traefik
Events:            <none>

    Curl attempts to 192.168.77.199:80 responds successfully with the desired Thank you for using nginx message.
    The results of ip a are as follows

    1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: ens160: mtu 1500 qdisc mq state UP group default qlen 1000 link/ether 00:0c:29:77:83:0b brd ff:ff:ff:ff:ff:ff inet 192.168.1.64/24 brd 192.168.1.255 scope global dynamic ens160 valid_lft 77141sec preferred_lft 77141sec inet6 fe80::20c:29ff:fe77:830b/64 scope link valid_lft forever preferred_lft forever 3: docker0: mtu 1500 qdisc noqueue state DOWN group default link/ether 02:42:54:a1:84:b8 brd ff:ff:ff:ff:ff:ff inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0 valid_lft forever preferred_lft forever 6: vxlan.calico: mtu 1440 qdisc noqueue state UNKNOWN group default link/ether 66:b0:66:fc:8e:8c brd ff:ff:ff:ff:ff:ff inet 192.168.153.64/32 brd 192.168.153.64 scope global vxlan.calico valid_lft forever preferred_lft forever inet6 fe80::64b0:66ff:fefc:8e8c/64 scope link valid_lft forever preferred_lft forever 7: [email protected]: mtu 1440 qdisc noqueue state UP group default link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 0 inet6 fe80::ecee:eeff:feee:eeee/64 scope link valid_lft forever preferred_lft forever 8: [email protected]: mtu 1440 qdisc noqueue state UP group default link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 1 inet6 fe80::ecee:eeff:feee:eeee/64 scope link valid_lft forever preferred_lft forever

    nc -vz requests to any of the IP's above at port 80 provide (tcp) failed: Connection refused
    The curl command as follows fails:

    curl -H "HOST: www.example.com" http://192.168.1.64:80 curl: (7) Failed to connect to 192.168.1.64 port 80: Connection refused
  • nkerr
    nkerr Posts: 10

    Hello,
    I seem to have a similar issue. I can curl the ip shown in ingress-test with success

    Name:             ingress-test
Namespace:        anthill
Address:          
Default backend:  default-http-backend:80 (<error: endpoints "default-http-backend" not found>)
Rules:
  Host             Path  Backends
  ----             ----  --------
  www.example.com  
                   /   secondapp:80   192.168.77.199:80)
Annotations:       kubernetes.io/ingress.class: traefik
Events:            <none>

    curl -H "HOST: www.example.com" http://192.168.77.199
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
...

    However I'm not sure if this is the correct IP. The following command doesn't provide any usable IP address for the ingress

    [email protected]:~$ ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
2: ens160: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:0c:29:77:83:0b brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.64/24 brd 192.168.1.255 scope global dynamic ens160
       valid_lft 76414sec preferred_lft 76414sec
    inet6 fe80::20c:29ff:fe77:830b/64 scope link 
       valid_lft forever preferred_lft forever
3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default 
    link/ether 02:42:54:a1:84:b8 brd ff:ff:ff:ff:ff:ff
    inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0
       valid_lft forever preferred_lft forever
6: vxlan.calico: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1440 qdisc noqueue state UNKNOWN group default 
    link/ether 66:b0:66:fc:8e:8c brd ff:ff:ff:ff:ff:ff
    inet 192.168.153.64/32 brd 192.168.153.64 scope global vxlan.calico
       valid_lft forever preferred_lft forever
    inet6 fe80::64b0:66ff:fefc:8e8c/64 scope link 
       valid_lft forever preferred_lft forever
7: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1440 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
8: ca[email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1440 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 1
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever

    Using the curl command with one of the above IP's doesn't work. nc commands to these result in (tcp) failed: Connection refused

    Not sure what I'm missing here.

  • chrispokorni
    chrispokorni Posts: 1,842

    Hi @nkerr,

    The ip a command is returning the node's IP address, that you can then test with curl -H .... If unsure, you may be able to confirm the node IP from your hypervisor or cloud console used to configure your node instances.

    From your output, however, it may seem that you cluster is misconfigured - your node IP addresses overlapping with the default pod network 192.168.0.0/16 managed by Calico. These two networks should not overlap in a cluster.

    Regards,
    -Chris

  • chrispokorni
    chrispokorni Posts: 1,842

    Hi @mmckenzie,

    Being able to access a pod, its service, and an ingress from only from the node where the pod is deployed, only shows that the networking is misconfigured at the cluster infrastructure level. It is not a solution, nor does it fix the networking issue.

    Regards,
    -Chris

  • nkerr
    nkerr Posts: 10
    edited May 2021

    @chrispokorni
    I'm not sure what indicator I've provided shows an overlapping IP addresses, but the kubernetes/calico cluster is configured based on the provided scripts with a slight alteration that was required to get it to work, Switching from IP-in-IP to VXLAN.

  • serewicz
    serewicz Posts: 1,000

    Hello,

    If you are using the 192.168 network for you nodes it will overlap with the default Kubernetes network range set via Calico and when the cluster is initialized. Rebuild you nodes to use a different IP range would be the easiest way to avoid this conflict.

    Regards,

Categories

Upcoming Training