Cluster agent in endless crash loop

SUSE Rancher Prime
1

Hi,

any idea how to investigate the root cause of the following error?
It’s a user cluster registered as “Custom” RKE cluster in Rancher HA 2.3.5
Nodes are created externally, no provider like AWS etc. used.

image
image889×394 18.1 KB

After this message nothing else follows and the cluster agent crashes.

image
image944×102 13.9 KB

Setup:

Rancher HA Cluster: v2.3.5
Nginx as external loadbalancer (in same network)
User cluster: rancher/rancher-agent:v2.3.5
Both Kubernetes v1.16.6-rancher1-2
Everything in same network

Any help appreciated.

Kind regards,
Michael

2

Installed the user cluster again, without to change the cidrs… now all pods in 10.42.0.0/16 subnet and services in 10.43.0.0/16…
The error remains same with different IP:

level=fatal msg=“Get https://10.43.0.1:443/apis/apiextensions.k8s.io/v1beta1/customresourcedefinitions: dial tcp 10.43.0.1:443: i/o timeout”

Did the DNS checks from here: https://rancher.com/docs/rancher/v2.x/en/troubleshooting/dns/
There are no errors in coredns, upstream dns is reachable too from all nodes (in this example only one control and one worker)

cattle-node-agents work fine.
Can’t find the reason. :grimacing:

3

OK, replaced Weave CNI trough Canal and it works… but I would prefer Weave because of encryption.

This failes as mentioned above

network:
    options:
      flannel_backend_type: vxlan
      plugin: weave
        weave_network_provider:
          password: ...

Canal works directly

network:
    options:
      flannel_backend_type: vxlan
      plugin: canal

Its a very basic setup with one control plane and one worker to evaluate the whole automated setup.
(https://rancher.com/docs/rke/latest/en/config-options/add-ons/network-plugins/)

Kind regards,
Michael