Magnum is an OpenStack API service developed by the OpenStack Containers Team making container orchestration engines (COE) such as Docker Swarm, Kubernetes and Apache Mesos available as the first class resources in OpenStack.


  • You need to have the heat_stack_owner role assigned to create Magnum clusters.

Create a cluster template

You need a Glance image that is supported by Magnum. The os_distro property has to be set properly on the image. We recommend the Fedora Atomic image, with the os_distro property set to fedora-atomic.

$ openstack coe cluster template create \
    --coe kubernetes \
    --dns-nameserver \
    --docker-volume-size 10 \
    --external-network public \
    --flavor 2C-16GB-20GB \
    --image 'Fedora Atomic Host 29' \
    --labels docker_volume_type=ceph-1,prometheus_monitoring=True \
    --master-flavor 1C-1GB-10GB \
    --master-lb-enabled \
    --network-driver flannel \
    'Fedora Atomic Host 29 - Kubernetes'


For additional parameters have a look at the Magnum user documentation:


To avoid problems with the Docker Hub, a Docker Registry Cache provided in the Betacloud can be used.

--labels [...],

Create a cluster

$ openstack coe cluster template list
| uuid                                 | name                               |
| 52edf7fb-d61a-4311-a99e-6d2f5c005a03 | Fedora Atomic Host 29 - Kubernetes |
$ openstack coe cluster create \
    --cluster-template 'Fedora Atomic Host 29 - Kubernetes' \
    --keypair KEYPAIR \
    --master-count 1 \
    --node-count 2 \
Request to create cluster 2548f2b3-64bd-487f-b5b5-003e631c8909 accepted


  • Save cluster config to a local directory and add it to environment

    $ mkdir magnum-testing
    $ openstack coe cluster config --dir magnum-testing testing
    $ export KUBECONFIG=magnum-testing/config
  • Now you can use kubectl as usually.

    $ kubectl cluster-info
    Kubernetes master is running at https://a.b.c.d:6443
    Heapster is running at https://a.b.c.d:6443/api/v1/namespaces/kube-system/services/heapster/proxy
    CoreDNS is running at https://a.b.c.d:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy
    To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.


  • If you receive HTTP 400 erros on cluster creation, try to use the magnum command instead of the openstack command.

  • Investigate your cluster

    $ openstack coe cluster list
    | uuid                                 | name    | keypair | node_count | master_count | status          |
    | 3ed5bc1d-6b08-4b58-ac7d-1410027ea574 | testing | KEYPAIR | 2          | 1            | CREATE_COMPLETE |
    $ openstack coe cluster -show <cluster>
  • Investigate the Heat stack of your cluster

    $ openstack coe cluster show testing -c stack_id
    $ openstack stack show <stack_id>
    $ openstack stack event list <stack_id>
  • Investigate the virtual machines of your cluster

    $ openstack stack resource list <stack_id> -n 2 \
        --filter type=OS::Nova::Server
    $ openstack console log show <server>