Kubernetes

Catalyst Kubernetes Service makes it easy for you to deploy, manage, and scale Kubernetes clusters to run containerised applications on the Catalyst Cloud.

Warning

This service is in Technical Preview (ALPHA) and is not recommended for production workloads yet.

Table of Contents:

Providing feedback

Our goal with this alpha release is to establish a feedback loop and collaborate with early adopters of the technology, to ensure it meets the unique needs of our customers in NZ.

At this stage, the service is expected to have some rough edges and bugs. If you encounter an issue or have a suggestion on how we can improve it, please raise a ticket via the Support Centre.

Where possible, when creating support tickets, please include the output of the following command to assist our support team in helping you to resolve it.

$ openstack coe cluster show <cluster_name>

Known Issues

This is an overview of the issues encountered in the process of testing this Technical Preview release.

Nodes lose their IP addresses

Description

This issue initially appears to be related to DNS as the error messages encountered when running kubectl will normally contain something similar to the following..

error getting address for node <nodename>: no address found for host

The actual problem however is related to the fact that the InternalIP and/or ExternalIP addresses are lost for some nodes.

Status

The cause of the problem has been identified and a fix has been proposed.

Workaround

Run docker restart for the kubelet container.

Cluster fails to delete

When deleting a cluster that has a loadbalancer added subsequent to the actual cluster deployment action, it will fail due to a resource conflict.

Failed to delete cluster because stack delete failed Error: Resource DELETE
failed: Conflict: resources.network.resources.private_subnet: Unable to
complete operation on subnet 0b65ff86-13a5-460f-96c3-d3b20377df60.
One or more ports have an IP allocation from this subnet.

Status

The cause of the problem has been identified and a fix is being investigated.

Workaround

Manually delete any loadbalancers within the cluster that were not created as part of the initial cluster creation.

Cluster is created but pods fail

The cluster create command completes and when examined appears to have done so successfully. Any attempts to interact with the nodes will fail.

If the current pod state is displayed it can be seen that there are several containers still sitting in the ContainerCreating state.

This is not a recoverable error and the cluster will need to be created if this occurs.

$ kubectl get pods --all-namespaces -o wide
NAMESPACE     NAME                                                   READY     STATUS              RESTARTS   AGE       IP          NODE                                     NOMINATED NODE
kube-system   ==calico-kube-controllers-54dfc58c64-2xprd==               1/1       Running             0          32m       10.0.0.13   test-prod-z2h4cs73basz-minion-0   <none>
kube-system   calico-node-95d26                                      2/2       Running             0          32m       10.0.0.12   test-prod-z2h4cs73basz-master-1   <none>
kube-system   calico-node-hkdzm                                      2/2       Running             0          28m       10.0.0.13   test-prod-z2h4cs73basz-minion-0   <none>
kube-system   calico-node-ljmxs                                      2/2       Running             0          32m       10.0.0.10   test-prod-z2h4cs73basz-master-0   <none>
kube-system   calico-node-vdczb                                      2/2       Running             0          32m       10.0.0.11   test-prod-z2h4cs73basz-master-2   <none>
kube-system   coredns-78df4bf8ff-kp9pk                               0/1       ContainerCreating   0          33m       <none>      test-prod-z2h4cs73basz-minion-0   <none>
kube-system   heapster-74f98f6489-7ctpn                              0/1       ContainerCreating   0          32m       <none>      test-prod-z2h4cs73basz-minion-0   <none>
kube-system   kube-dns-autoscaler-986c49747-whkmh                    0/1       ContainerCreating   0          33m       <none>      test-prod-z2h4cs73basz-minion-0   <none>
kube-system   kubernetes-dashboard-54cb7b5997-2hzzk                  0/1       ContainerCreating   0          32m       <none>      test-prod-z2h4cs73basz-minion-0   <none>
kube-system   node-exporter-test-prod-z2h4cs73basz-minion-0          0/1       ContainerCreating   0          23m       <none>      test-prod-z2h4cs73basz-minion-0   <none>

Status

The cause of the problem has been identified and a fix is being investigated.

Workaround

There is currently an interim workaround in place to address the issue.

Cluster takes a long time to deploy

Description

Currently the time taken to deploy a cluster from commandline or dashbard is in the vicinity of 15-25 minutes.

Status

The cause of the problem is known and a fix is being investigated.

Workaround

None.