Add flannel error

[wyfligamo]

quickstart version: beta-kubernetes:0.4.0-1.9.0-beta
dcos version: 1.10.4

I installed kubernetes with web ui,it seems every service in dcos thing is going on and I can get three Ready nodes

[root@localhost ~]# kubectl get nodes
NAME                                   STATUS    ROLES     AGE       VERSION
kube-node-0-kubelet.kubernetes.mesos   Ready         23h       v1.9.0
kube-node-1-kubelet.kubernetes.mesos   Ready         23h       v1.9.0
kube-node-2-kubelet.kubernetes.mesos   Ready         23h       v1.9.0

Then I try to add flannel to cluster, but get an error

[root@localhost ~]# kubectl get pods --namespace=kube-system
NAME                        READY     STATUS             RESTARTS   AGE
heapster-7d7b7f4f87-j4j4j   1/1       Running            0          23h
kube-dns-6dc79b7c4-kgmc2    2/3       CrashLoopBackOff   555        23h
kube-flannel-ds-9mkw4       0/1       CrashLoopBackOff   4          2m
kube-flannel-ds-s7w6h       0/1       CrashLoopBackOff   4          2m
kube-flannel-ds-xncpc       0/1       CrashLoopBackOff   4          2m

Logs look like this

[root@localhost ~]# kubectl logs --namespace=kube-system kube-flannel-ds-9mkw4 
I0201 02:24:56.214697       1 main.go:474] Determining IP address of default interface
I0201 02:24:56.218057       1 main.go:487] Using interface with name eth0 and address 10.100.100.15
I0201 02:24:56.218141       1 main.go:504] Defaulting external address to interface address (10.100.100.15)
I0201 02:24:56.277590       1 kube.go:130] Waiting 10m0s for node controller to sync
I0201 02:24:56.277757       1 kube.go:283] Starting kube subnet manager
I0201 02:24:57.277999       1 kube.go:137] Node controller sync successful
I0201 02:24:57.278089       1 main.go:234] Created subnet manager: Kubernetes Subnet Manager - kube-node-1-kubelet.kubernetes.mesos
I0201 02:24:57.278107       1 main.go:237] Installing signal handlers
I0201 02:24:57.278338       1 main.go:352] Found network config - Backend type: vxlan
I0201 02:24:57.278490       1 vxlan.go:119] VXLAN config: VNI=1 Port=0 GBP=false DirectRouting=false
E0201 02:24:57.279419       1 main.go:279] Error registering network: failed to acquire lease: node "kube-node-1-kubelet.kubernetes.mesos" pod cidr not assigned
I0201 02:24:57.279476       1 main.go:332] Stopping shutdownHandler...

How can I fix this?

[hectorj2f]

@wyflimago Why do you need to install flannel ?

[pires]

Hi @wyfligamo! Currently, we support only the DC/OS overlay. We have been trying some other overlay networks, like Calico, but it will need more work, and so we'll leave it to post-GA.

Adding some more details, in this particular case, the error is:

E0201 02:24:57.279419 1 main.go:279] Error registering network: failed to acquire lease: node "kube-node-1-kubelet.kubernetes.mesos" pod cidr not assigned
This happens because we do not enable the --allocate-node-cidrs flag for the kube-controller-manager.

[pires]

@wyfligamo feel free to comment but I'm going to close this as it's no bug. Thank you for reporting.

[wyfligamo]

@hectorj2f my kube-dns pod status is CrashLoopBackOff and kubedns container log says

E0202 01:21:34.174837       1 reflector.go:201] k8s.io/dns/pkg/dns/dns.go:147: Failed to list *v1.Endpoints: Get https://10.100.0.1:443/api/v1/endpoints?resourceVersion=0: dial tcp 10.100.0.1:443: i/o timeout
E0202 01:21:34.175333       1 reflector.go:201] k8s.io/dns/pkg/dns/dns.go:150: Failed to list *v1.Service: Get https://10.100.0.1:443/api/v1/services?resourceVersion=0: dial tcp 10.100.0.1:443: i/o timeout

I can't find any other pods except heapster

[root@localhost ~]# kubectl get pods --all-namespaces 
NAMESPACE     NAME                        READY     STATUS             RESTARTS   AGE
kube-system   heapster-7d7b7f4f87-brp2l   1/1       Running            0          21h
kube-system   kube-dns-6dc79b7c4-dbw5g    1/3       CrashLoopBackOff   506        21h

so I wonder if I should add a network plugin like flannel

[wyfligamo]

@pires I don't understand about DC/OS overlay, is that working as a kubernetes pods or a dcos/k8s service?
My k8s only start kube-dns and heapster pods, is that right? they can't communicate with api service, but I can get api response with browser outside the cluster

[root@localhost ~]# kubectl logs --namespace=kube-system heapster-7d7b7f4f87-brp2l
E0202 05:26:09.524966       1 reflector.go:190] k8s.io/heapster/metrics/heapster.go:322: Failed to list *v1.Pod: Get https://kubernetes.default/api/v1/pods?resourceVersion=0: dial tcp: i/o timeout

[root@localhost ~]# kubectl logs --namespace=kube-system kube-dns-6dc79b7c4-dbw5g kubedns
E0202 05:23:47.244936       1 reflector.go:201] k8s.io/dns/pkg/dns/dns.go:150: Failed to list *v1.Service: Get https://10.100.0.1:443/api/v1/services?resourceVersion=0: dial tcp 10.100.0.1:443: i/o timeout
E0202 05:23:47.244981       1 reflector.go:201] k8s.io/dns/pkg/dns/dns.go:147: Failed to list *v1.Endpoints: Get https://10.100.0.1:443/api/v1/endpoints?resourceVersion=0: dial tcp 10.100.0.1:443: i/o timeout
I0202 05:23:47.742922       1 dns.go:173] Waiting for services and endpoints to be initialized from apiserver...
I0202 05:23:48.242824       1 dns.go:173] Waiting for services and endpoints to be initialized from apiserver...
I0202 05:23:48.742850       1 dns.go:173] Waiting for services and endpoints to be initialized from apiserver...
I0202 05:23:49.242879       1 dns.go:173] Waiting for services and endpoints to be initialized from apiserver...

[pires]

@wyfligamo https://docs.mesosphere.com/1.10/networking/virtual-networks/#dcos-overlay

[pires]

Since this issue was initially about running flannel and that has been clarified to the user, I am moving the real issue to a new issue #58. Let's continue there.