[magnum] Can not deploy a cluster using Fedora CoreOS 32
I am trying to use Magnum on Openstack that was deployed using kolla-ansible. The instance is completely private (so floating IPs are only accessible inside a VPN network) - not sure if this is a problem. The deployment looks as follows: Openstack Nodes: Ubuntu 20.04 Kolla-Ansible version: '10.1.0' kolla_base_distro: "ubuntu" openstack_release: "ussuri" I could successfully deploy the required magnum services and created a cluster template using the magnum cli as follows: openstack coe cluster template create kubernetes-cluster-template --image fedora-coreos-32 --external-network public-external --dns-nameserver 1.1.1.1 --master-flavor m1.small --flavor m1.small --docker-storage-driver overlay --coe kubernetes I am using the Fedora CoreOS 32 release (and uploaded the image like this): openstack image create --disk-format=qcow2 --container-format=bare --file=fedora-coreos-32.20201104.3.0-openstack.x86_64.qcow2 --property os_distro='fedora-coreos' fedora-coreos-32 I made sure to set the os_distro as stated in the repository when using fedora coreos. When I now try to create a cluster (openstack coe cluster create kubernetes-cluster --cluster-template kubernetes-cluster-template --master-count 1 --node-count 1 --keypair mykey) all the heat resources start spinning up, but the kube-masters resource will never come up. While debugging it seems to hang during the following step inside the heat-container-agent: + '[' ok = ok ']' + kubectl patch node kubernetes-cluster-flfrufngma6x-master-0 --patch '{"metadata": {"labels": {"node-role.kubernetes.io/master": ""}}}' error: no configuration has been provided, try setting KUBERNETES_MASTER environment variable + echo 'Trying to label master node with node-role.kubernetes.io/master=""' + sleep 5s Trying to label master node with node-role.kubernetes.io/master="" I also didn't find a way to set the KUBERNETES_MASTER environment variable from the outside. I assume there is no other supported distribution given that Fedora Atomic and non-Fedora CoreOS do no longer exist? Sent with [ProtonMail](https://protonmail.com) Secure Email.
Try https://hub.docker.com/layers/openstackmagnum/heat-container-agent/ussuri-st... For heat_container_agent_tag Spyros On Wed, Nov 25, 2020 at 4:48 PM monadt <monadt@protonmail.com> wrote:
I am trying to use Magnum on Openstack that was deployed using kolla-ansible.
The instance is completely private (so floating IPs are only accessible inside a VPN network) - not sure if this is a problem.
The deployment looks as follows:
Openstack Nodes: Ubuntu 20.04 Kolla-Ansible version: '10.1.0' kolla_base_distro: "ubuntu" openstack_release: "ussuri"
I could successfully deploy the required magnum services and created a cluster template using the magnum cli as follows:
openstack coe cluster template create kubernetes-cluster-template --image fedora-coreos-32 --external-network public-external --dns-nameserver 1.1.1.1 --master-flavor m1.small --flavor m1.small --docker-storage-driver overlay --coe kubernetes
I am using the Fedora CoreOS 32 release (and uploaded the image like this):
openstack image create --disk-format=qcow2 --container-format=bare --file=fedora-coreos-32.20201104.3.0-openstack.x86_64.qcow2 --property os_distro='fedora-coreos' fedora-coreos-32
I made sure to set the os_distro as stated in the repository when using fedora coreos.
When I now try to create a cluster (openstack coe cluster create kubernetes-cluster --cluster-template kubernetes-cluster-template --master-count 1 --node-count 1 --keypair mykey) all the heat resources start spinning up, but the kube-masters resource will never come up.
While debugging it seems to hang during the following step inside the heat-container-agent:
+ '[' ok = ok ']' + kubectl patch node kubernetes-cluster-flfrufngma6x-master-0 --patch '{"metadata": {"labels": {"node-role.kubernetes.io/master": ""}}}' error: no configuration has been provided, try setting KUBERNETES_MASTER environment variable + echo 'Trying to label master node with node-role.kubernetes.io/master= ""' + sleep 5s Trying to label master node with node-role.kubernetes.io/master=""
I also didn't find a way to set the KUBERNETES_MASTER environment variable from the outside.
I assume there is no other supported distribution given that Fedora Atomic and non-Fedora CoreOS do no longer exist?
Sent with ProtonMail <https://protonmail.com> Secure Email.
Monadt, I proposed some basic docs for using magnum with Kolla Ansible [1]. They don't cover the cluster template, but they might be useful to you anyway. [1] https://review.opendev.org/c/openstack/kolla-ansible/+/763770 On Wed, 25 Nov 2020 at 15:49, monadt <monadt@protonmail.com> wrote:
I am trying to use Magnum on Openstack that was deployed using kolla-ansible.
The instance is completely private (so floating IPs are only accessible inside a VPN network) - not sure if this is a problem.
The deployment looks as follows:
Openstack Nodes: Ubuntu 20.04 Kolla-Ansible version: '10.1.0' kolla_base_distro: "ubuntu" openstack_release: "ussuri"
I could successfully deploy the required magnum services and created a cluster template using the magnum cli as follows:
openstack coe cluster template create kubernetes-cluster-template --image fedora-coreos-32 --external-network public-external --dns-nameserver 1.1.1.1 --master-flavor m1.small --flavor m1.small --docker-storage-driver overlay --coe kubernetes
I am using the Fedora CoreOS 32 release (and uploaded the image like this):
openstack image create --disk-format=qcow2 --container-format=bare --file=fedora-coreos-32.20201104.3.0-openstack.x86_64.qcow2 --property os_distro='fedora-coreos' fedora-coreos-32
I made sure to set the os_distro as stated in the repository when using fedora coreos.
When I now try to create a cluster (openstack coe cluster create kubernetes-cluster --cluster-template kubernetes-cluster-template --master-count 1 --node-count 1 --keypair mykey) all the heat resources start spinning up, but the kube-masters resource will never come up.
While debugging it seems to hang during the following step inside the heat-container-agent:
+ '[' ok = ok ']' + kubectl patch node kubernetes-cluster-flfrufngma6x-master-0 --patch '{"metadata": {"labels": {"node-role.kubernetes.io/master": ""}}}' error: no configuration has been provided, try setting KUBERNETES_MASTER environment variable + echo 'Trying to label master node with node-role.kubernetes.io/master=""' + sleep 5s Trying to label master node with node-role.kubernetes.io/master=""
I also didn't find a way to set the KUBERNETES_MASTER environment variable from the outside.
I assume there is no other supported distribution given that Fedora Atomic and non-Fedora CoreOS do no longer exist?
Sent with ProtonMail Secure Email.
participants (3)
-
Mark Goddard
-
monadt
-
Spyros Trigazis