Liqo installation Problem

[wefersmasasana]

Hello there,

I have a problem with Liqo I wanted to install the helm chart with the values. On my provider server (Server1) this was not a problem. All pods and alls services are running. But on my consumer server (Server2) when I tried to install with the values, this error came up:

INSTALLATION FAILED: failed post-install: warning: Hook post-install liqo/templates/webhooks/job-patch/rbac.yaml failed: 1 error occurred:
* Post "https://5.161.118.255:16443/apis/rbac.authorization.k8s.io/v1/clusterrolebindings?fieldManager=helm": read tcp 5.161.118.255:40436->5.161.118.255:16443: read: connection reset by peer

But the installation continues and the pods are running, but two services are pending.
In attachment of this mail are included screenshots when I executed „Liqoctl status“ on both servers and screenshots of the services running in Kubernetes.

Thanks for your help!

Yours faithfully
Marcel Wefers

Server 1

Server 2

[aleoli]

Hi @wefersmasasana! In my opinion, you have two different problems:

1. the LoadBalancer services are pending, probably because you don't have any lb provider in the second cluster. You can either install one (as MetalLB) or use NodePort services
2. a connection error from Helm to your API server during installation. The port looks strange to me (16443 instead of 6443), but it is ok if the previous steps run correctly. If you try to install it again, do you know if this problem persists?

[wefersmasasana]

Thank your for your answer, I installed a load balancer (MetalLB) now while installing this error:

"INSTALLATION FAILED: failed post-install: warning: Hook post-install liqo/templates/webhooks/job-patch/rbac.yaml failed: 1 error occurred:

• Post "https://5.161.118.255:16443/apis/rbac.authorization.k8s.io/v1/clusterrolebindings?fieldManager=helm": read tcp 5.161.118.255:40436->5.161.118.255:16443: read: connection reset by peer"

is gone but the services and the Liqo status are the same as in the screenshots I sended.

[aleoli]

Ok, is MetalLB creating some events on these services or logging something?

Also, if you create another LB service, does it work? If not, it should be a MetalLB-related problem, not a Liqo one.

[wefersmasasana]

Thank you Liqo is now running on my second cluster the problem was that no loadbalancer was installed. I installed MetalLB, but now there is another problem. When I generate my peer command and copy it and paste it to my second cluster this error comes up:

[aleoli]

You probably have an old foreign cluster from a previous installation, try deleting it and re-try the peering

kubectl delete foreigncluster marcel

[wefersmasasana]

[aleoli]

The validation error is due to an update of a foreigncluster resource; if no one exists, it should work. Did you check on both clusters?

[wefersmasasana]

I needed to delete the foreign cluster on the second Server but now this error came up after 2minuted trying to peer:

[aleoli]

Can you check the state of foreign clusters? You will probably have broken authentication and/or pending peering. You can try unpeer and then delete both liqo-tenant-* namespaces to clean up the dirty state. If this is a testing environment, I suggest you to delete and re-create your clusters to be sure everything is clean.

[wefersmasasana]

Hey again. I had deleted my clusters and created new ones. When I try to peer the clusters again, the same error came up:

"ERROR Authentication to the remote cluster "marcel" failed: timed out waiting for the condition"

[wefersmasasana]

[aleoli]

Can you check the logs of your liqo controller manager? Additionally, can you check that your LoadBalancer services are reachable from the other cluster?

[wefersmasasana]

Hi sorry for not responding. This are the logs from the controller manager pod:

[aleoli]

You probably have a dirty secret in your liqo or liqo-tenant namespace, try deleting all remote token secrets