Skip to content

Latest commit

 

History

History
67 lines (51 loc) · 3.16 KB

trouble_shooting.md

File metadata and controls

67 lines (51 loc) · 3.16 KB

Table of Contents generated with DocToc

Trouble shooting

This guide (based on minikube and others should be similar) explains general info on how to debug issues if cluster failed to create.

Get log of clusterapi-controllers containers

  1. Get openstack container name, the output depends on the system you are running. the minikube.kubeconfig which is bootstrap cluster's kubeconfig by default locates at cmd/clusterctl folder.

    # kubectl --kubeconfig minikube.kubeconfig get pods -n openstack-provider-system
    NAMESPACE                   NAME                                     READY   STATUS    RESTARTS   AGE
    openstack-provider-system   clusterapi-controllers-xxxxxxxxx-xxxxx   1/1     Running   0          27m
    
  2. Get log of clusterapi-controllers-xxxxxxxx-xxxxx

    # kubectl --kubeconfig minikube.kubeconfig log clusterapi-controllers-xxxxxxxxx-xxxxx -n openstack-provider-system
    

Master failed to start with error: node xxxx not found

Sometimes the master machine is created but failed to startup, take ubuntu as example, open /var/log/messages and if you see something like

Jul 10 00:07:58 openstack-master-5wgrw kubelet: E0710 00:07:58.444950 4340 kubelet.go:2248] node "openstack-master-5wgrw" not found
Jul 10 00:07:58 openstack-master-5wgrw kubelet: I0710 00:07:58.526091 4340 kubelet_node_status.go:72] Attempting to register node openstack-master-5wgrw
Jul 10 00:07:58 openstack-master-5wgrw kubelet: E0710 00:07:58.527398 4340 kubelet_node_status.go:94] Unable to register node "openstack-master-5wgrw" with API server: nodes "openstack-master-5wgrw" is forbidden: node "openstack-master-5wgrw.novalocal" is not allowed to modify node "openstack-master-5wgrw"

This might be caused by This issue, try the method proposed there.

providerClient authentication err

If you are using https, you must specify the cacert in your clouds.yaml file, and when you encounter issue like:

# kubectl logs clusterapi-controllers-0 -n openstack-provider-system
...
E0814 04:32:52.688514       1 machine_controller.go:204] Failed to check if machine "openstack-master-hxk9r" exists: providerClient authentication err: Post https://xxxxxxxxxxxxxxx:5000/v3/auth/tokens: x509: certificate signed by unknown authority
...

you can add verify: false into clouds.yaml file to solve the problem.

clouds:
  openstack:
    auth:
        ....
    region_name: "RegionOne"
    interface: "public"
    identity_api_version: 3
    cacert: /etc/certs/cacert
    verify: false