Getting error in agent-clusternet in child clusternet

[shekhar250723]

I am getting error in child-clusternet setup in GCP account.

Case 1- On my laptop
Getting below error in my GCP account.
Error: Kubernetes cluster unreachable: Get "https://104.197.186.74/version?timeout=32s": x509: certificate has expired or is not yet valid: current time 2023-08-30T17:38:06+05:30 is before 2023-08-30T12:45:29Z

Created new token but still getting error.

Note:- Parent cluster has been created in different project and child clusternet has been setup in different project.

Case 2- Different laptop
Getting error in child clusternet whenever tried to install agent-clusternet under of this. We are using child clusternet in different gcp and parent clusternet in different gcp account.

Agent-clusternet has been installed but getting below error in pod status

Error
kubectl get pod -n clusternet-system
NAME READY STATUS RESTARTS AGE
clusternet-agent-99d96988b-kclql 0/1 Error 2 (19s ago) 27s
clusternet-agent-99d96988b-kqd7x 0/1 Error 2 (19s ago) 27s
clusternet-agent-99d96988b-w7cjf 0/1 Error 2 (19s ago) 27s

Please help me on this.

[dixudx]

Error: Kubernetes cluster unreachable: Get "https://104.197.186.74/version?timeout=32s": x509: certificate has expired or is not yet valid: current time 2023-08-30T17:38:06+05:30 is before 2023-08-30T12:45:29Z

Actually it is NOT a problem of Clusternet.

The certificates of Kubernetes clusters got expired. You can try multiple ways to fix the expired certificates, such as

• https://kubernetes.io/docs/tasks/administer-cluster/kubeadm/kubeadm-certs/
• https://kubernetes.io/docs/tasks/tls/managing-tls-in-a-cluster/
• ...

[shekhar250723]

@dixudx It has been resolved automatically after the re-installation.

But I have some doubts. I have created one parent and 2 chilld cluster but whenever I tried to check the status of first child cluster through parent cluster then it's working fine but I got same status in another child cluster.

Second if I tried to switched the child cluster directly. Then I didn't check the created ns which I have created through parent cluster and getting error.

Error 1- Created the ns in first child cluster but it's also visible in another child cluster

$ kubectl clusternet --cluster-id=407df4f8-09f3-4353-bd13-0bbe23374669 --kubeconfig=.kube/config create ns developer-namespace
namespace/developer-namespace created

$ kubectl get clsrr
NAME                                              CLUSTER ID                             STATUS     AGE
clusternet-3ee7dab1-02c5-458d-8f3a-9e48d30af202   3ee7dab1-02c5-458d-8f3a-9e48d30af202   Approved   10m
clusternet-407df4f8-09f3-4353-bd13-0bbe23374669   407df4f8-09f3-4353-bd13-0bbe23374669   Approved   16h

$ kubectl clusternet --cluster-id=407df4f8-09f3-4353-bd13-0bbe23374669 --kubeconfig=.kube/config get ns
NAME                  STATUS   AGE
clusternet-9gw4k      Active   10m
clusternet-reserved   Active   16h
clusternet-system     Active   16h
clusternet-szc2b      Active   16h
default               Active   16h
developer-namespace   Active   19s
gmp-public            Active   16h
gmp-system            Active   16h
kube-node-lease       Active   16h
kube-public           Active   16h
kube-system           Active   16h

$ kubectl clusternet --cluster-id=3ee7dab1-02c5-458d-8f3a-9e48d30af202 --kubeconfig=.kube/config get ns
NAME                  STATUS   AGE
clusternet-9gw4k      Active   11m
clusternet-reserved   Active   16h
clusternet-system     Active   16h
clusternet-szc2b      Active   16h
default               Active   16h
developer-namespace   Active   26s
gmp-public            Active   16h
gmp-system            Active   16h
kube-node-lease       Active   16h
kube-public           Active   16h
kube-system           Active   16h 

Error 2- Getting error if I switched directly the child cluster and tried to check the created ns which is done by parent clusternet command.

$ kubectl clusternet get ns
Error from server (NotFound): Unable to list "/v1, Resource=namespaces": the server could not find the requested resource (get namespaces)

[dixudx]

For error 1

It seems that the same cluster is registered twice.

For error 2

You are not specifying a valid kubeconfig. By default $HOME/.kube/config is used.

[shekhar250723]

@dixudx As checked again, no same cluster has been registered twice. Could you please let me know any command? How can we check the status of it.

Second - Kubeconfig file is valid and updated the same as per the process but still getting the same error.

[dixudx]

$ kubectl clusternet --cluster-id=407df4f8-09f3-4353-bd13-0bbe23374669 --kubeconfig=.kube/config get ns
NAME                  STATUS   AGE
clusternet-9gw4k      Active   10m
clusternet-reserved   Active   16h
clusternet-system     Active   16h
clusternet-szc2b      Active   16h
default               Active   16h
developer-namespace   Active   19s
gmp-public            Active   16h
gmp-system            Active   16h
kube-node-lease       Active   16h
kube-public           Active   16h
kube-system           Active   16h

$ kubectl clusternet --cluster-id=3ee7dab1-02c5-458d-8f3a-9e48d30af202 --kubeconfig=.kube/config get ns
NAME                  STATUS   AGE
clusternet-9gw4k      Active   11m
clusternet-reserved   Active   16h
clusternet-system     Active   16h
clusternet-szc2b      Active   16h
default               Active   16h
developer-namespace   Active   26s
gmp-public            Active   16h
gmp-system            Active   16h
kube-node-lease       Active   16h
kube-public           Active   16h
kube-system           Active   16h 

They have exactly the same output.
@shekhar250723 Where is the child kubeconfig? I did not see any --child-kubeconfig flag here. Would you please follow the guide strictly? Thanks.

Moreover, flag --kubeconfig and --child-kubeconfig are different, they are for different clusters. You can tell them from the flag name.

[shekhar250723]

@dixudx Sorry for that, but it's not working and clusternet.io documentation is incomplete. Child config command is not working properly.

$kubectl clusternet --cluster-id=59a4d6ab-14de-473c-8e1d-500bc5cdbe81 --kubeconfig=.kube/config  get ns
NAME                  STATUS   AGE
clusternet-gpv86      Active   18h
clusternet-reserved   Active   18h
clusternet-system     Active   18h
default               Active   42h
gmp-public            Active   42h
gmp-system            Active   42h
kube-node-lease       Active   42h
kube-public           Active   42h
kube-system           Active   42h
dev-link               Active   126m

Error

$ kubectl clusternet --cluster-id=59a4d6ab-14de-473c-8e1d-500bc5cdbe81 --child-kubeconfig=.kube/config get ns
error: the server doesn't have a resource type "ns"

--child-kubeconfig command not working

[dixudx]

Firstly --child-kubeconfig and --kubeconfig are different flags. I am not meaning that you can only use either of them. Flag --kubeconfig will normally load from $HOME/.kube/config by default, unless you're explicitly specifying one.

If you're not clear about the default values, I'd suggest you adopt an explicit setting as follows,

$ kubectl clusternet --cluster-id=59a4d6ab-14de-473c-8e1d-500bc5cdbe81 --kubeconfig=[your parent cluster kubeconfig path]  --child-kubeconfig=[your child cluster kubeconfig path] get ns

Secondly, normally you should have different kubeconfig files for flag --kubeconfig and --child-kubeconfig respectively, unless you know exactly that you register parent cluster to itself as a child cluster.

Last but not the least, you can learn more from https://kubernetes.io/docs/concepts/configuration/organize-cluster-access-kubeconfig/, where you could have a better understanding of kubeconfig files, default locations, basic usages ,etc.

Currently what you've met now are solely Kubernetes usage problems. Please try to know exactly the command you're issuing.