Cluster provisioning "succeeds" despite machine-controller failing to start
rsdcastro opened this issue · 1 comments
Logs from cluster-api-gcp:
I0220 15:20:25.802792 19570 deploy_helper.go:68] Created master gce-master-test1
I0220 15:21:05.422987 19570 deploy_helper.go:241] Waiting for Kubernetes to come up...
I0220 15:21:13.538578 19570 deploy_helper.go:241] Waiting for Kubernetes to come up...
I0220 15:21:21.449076 19570 deploy_helper.go:241] Waiting for Kubernetes to come up...
I0220 15:21:29.529327 19570 deploy_helper.go:241] Waiting for Kubernetes to come up...
I0220 15:21:37.577458 19570 deploy_helper.go:274] wrote kubeconfig to [/Users/rdc/.kube/config]
I0220 15:21:37.577530 19570 deploy_helper.go:85] Waiting for apiserver to become healthy...
I0220 15:22:14.658776 19570 deploy_helper.go:94] Waiting for the service account to exist...
I0220 15:22:25.097828 19570 deploy_helper.go:99] Starting the machine controller...
I0220 15:22:46.073147 19570 deploy_helper.go:136] Clusters CRD created succuessfully!
I0220 15:22:46.849643 19570 deploy_helper.go:160] Machines CRD created successfully!
I0220 15:22:46.981945 19570 deploy_helper.go:176] Added machine [gce-master-test1]
I0220 15:22:47.088406 19570 deploy_helper.go:176] Added machine [gce-node-l71sp]
I0220 15:22:47.088443 19570 deploy.go:74] The [test1] cluster has been created successfully!
I0220 15:22:47.088455 19570 deploy.go:75] You can now kubectl get nodes. Note: you might only see the master for the first few minutes while the machine controller spins up the node machines.
But if one checks the machine controller logs:
$ kubectl logs machine-controller -n kube-system
F0220 23:34:12.548770 80 machinecontroller.go:71] error creating machine actuator: Not recognized cloud provider: terraform
This means that ideally here we can find a way to check if the machine controller is actually up & running to act on the machine objects before returning success to the user.
This issue was moved to kubernetes-sigs/cluster-api#58