Open mnaser opened 1 year ago
berendt
commented
1 year ago I started with the integration in Kolla-Ansible: https://review.opendev.org/c/openstack/kolla-ansible/+/896292.
With this patch and installed magnum-cluster-api plugin inside the magnum container images I have this error when trying to create the COE:
Cluster type (vm, Unset, kubernetes) not supported (HTTP 400) (Request-ID: req-71b9046a-61d2-498d-b0f4-243ce729f9cf)Wrong image ID. It works with a correct image ID:
Request to create cluster template k8s-v1.27.5 accepted
+-----------------------+--------------------------------------+
| Field | Value |
+-----------------------+--------------------------------------+
| insecure_registry | - |
| labels | {'kube_tag': 'v1.27.5'} |
| updated_at | - |
| floating_ip_enabled | True |
| fixed_subnet | - |
| master_flavor_id | SCS-2V-4-10 |
| uuid | df0e74af-6938-40a6-9062-2f9de3c316da |
| no_proxy | - |
| https_proxy | - |
| tls_disabled | False |
| keypair_id | - |
| public | False |
| http_proxy | - |
| docker_volume_size | - |
| server_type | vm |
| external_network_id | public |
| cluster_distro | ubuntu |
| image_id | 929b90b6-202a-4651-b3d2-2b70e39d592f |
| volume_driver | - |
| registry_enabled | False |
| docker_storage_driver | overlay2 |
| apiserver_port | - |
| name | k8s-v1.27.5 |
| created_at | 2023-09-22T20:16:16+00:00 |
| network_driver | calico |
| fixed_network | - |
| coe | kubernetes |
| flavor_id | SCS-2V-4-10 |
| master_lb_enabled | True |
| dns_nameserver | 8.8.8.8 |
| hidden | False |
| tags | - |
+-----------------------+--------------------------------------+After installing helm in the magnum-conductor container image + provisioning of Manila & Barbican I can start the creation of new clusters. I also have a magnum-system namespace in the Cluster API management cluster:
$ kubectl get namespaces
NAME STATUS AGE
default Active 2d14h
kube-system Active 2d14h
kube-public Active 2d14h
kube-node-lease Active 2d14h
metallb-system Active 2d14h
cert-manager Active 2d14h
capi-system Active 2d14h
capi-kubeadm-bootstrap-system Active 2d14h
capi-kubeadm-control-plane-system Active 2d14h
capo-system Active 2d14h
magnum-system Active 41mCreation of the cluster itself is hanging:
2023-09-22 21:07:21.165 7 DEBUG magnum.service.periodic [-] Updating status for cluster 9 update_status /var/lib/kolla/venv/lib/python3.10/site-packages/magnum/service/periodic.py:67
2023-09-22 21:07:21.257 7 DEBUG magnum.service.periodic [-] Status for cluster 9 updated to CREATE_IN_PROGRESS (CAPI Cluster status: Provisioning: Cluster kube-bpm9m is Provisioning. CAPI OpenstackCluster status reason: ) update_status /var/lib/kolla/venv/lib/python3.10/site-packages/magnum/service/periodic.py:84Maybe the used Cluster API & CAPO versions are too new?
CAPI_VERSION="v1.5.1"
CAPO_VERSION="v0.8.0"
clusterctl init \
--core cluster-api:${CAPI_VERSION} \
--bootstrap kubeadm:${CAPI_VERSION} \
--control-plane kubeadm:${CAPI_VERSION} \
--infrastructure openstack:${CAPO_VERSION}
mnaser
commented
1 year ago @berendt I noticed that 1.5 is failing to provision new cluster, I haven't gotten around the root cause, I think it might be fixed in the recent commits. 1.4.x works fine, but you'll find out about 1.5 here
mnaser
commented
1 year ago @berendt you should be good to go now with tip of branch, i'll release a new version soon with the fix :)
berendt
commented
1 year ago I haven't had time to continue yet. I will probably have time for the topic on Thursday.
berendt
commented
1 year ago @mnaser Is it possible to do the release today? We want to do some tests tomorrow and would be helpful to be able to work with the bug fix.
nguyenhuukhoi
commented
11 months ago Wrong image ID. It works with a correct image ID:
Request to create cluster template k8s-v1.27.5 accepted +-----------------------+--------------------------------------+ | Field | Value | +-----------------------+--------------------------------------+ | insecure_registry | - | | labels | {'kube_tag': 'v1.27.5'} | | updated_at | - | | floating_ip_enabled | True | | fixed_subnet | - | | master_flavor_id | SCS-2V-4-10 | | uuid | df0e74af-6938-40a6-9062-2f9de3c316da | | no_proxy | - | | https_proxy | - | | tls_disabled | False | | keypair_id | - | | public | False | | http_proxy | - | | docker_volume_size | - | | server_type | vm | | external_network_id | public | | cluster_distro | ubuntu | | image_id | 929b90b6-202a-4651-b3d2-2b70e39d592f | | volume_driver | - | | registry_enabled | False | | docker_storage_driver | overlay2 | | apiserver_port | - | | name | k8s-v1.27.5 | | created_at | 2023-09-22T20:16:16+00:00 | | network_driver | calico | | fixed_network | - | | coe | kubernetes | | flavor_id | SCS-2V-4-10 | | master_lb_enabled | True | | dns_nameserver | 8.8.8.8 | | hidden | False | | tags | - | +-----------------------+--------------------------------------+I started with the integration in Kolla-Ansible: https://review.opendev.org/c/openstack/kolla-ansible/+/896292.
With this patch and installed magnum-cluster-api plugin inside the magnum container images I have this error when trying to create the COE:
Cluster type (vm, Unset, kubernetes) not supported (HTTP 400) (Request-ID: req-71b9046a-61d2-498d-b0f4-243ce729f9cf)
I encountered this problem too. Can you share me solution for it. Thank you,
nguyenhuukhoi
commented
11 months ago Hello. my problem happened because magnum capi using image_id instead image name when we create cluster template.
nguyenhuukhoi
commented
11 months ago I added more details how make Magnum CAPI to work with Openstack
Requirements:
We setup a k8s cluster to manage workload clusters which were created by Magnum.
We need get /root/.kube/config from cluster management.
https://cluster-api.sigs.k8s.io/user/quick-start.html
We need log in(as root) to magnum api and magnum conductor and install. docker exec -u root 'magnum container' /bin/bash
apt update
pip install magnum-cluster-api
wget https://get.helm.sh/helm-v3.13.2-linux-amd64.tar.gz # We can choose other vervion
tar -zxvf helm-v3.13.2-linux-amd64.tar.gz
mv linux-amd64/helm /usr/local/bin/We need helm to make this work. Because There is no helm in magnum containers.
Copy /root/.kube/config from cluster management to magnum container under path /var/lib/magnum/.kube/config
I tested on running container directly. If you want to build images. You must do above steps when use kolla build. You need read followed link too.
https://review.opendev.org/c/openstack/kolla-ansible/+/896292
https://vexxhost.github.io/magnum-cluster-api/user/getting-started/
Feel free to ask me
mnaser
commented
10 months ago The progress to improve the Kolla support is in place:
kubeconfignguyenhuukhoi
commented
10 months ago The progress to improve the Kolla support is in place:
- https://review.opendev.org/c/openstack/kolla/+/902101 to add it to the images
- https://review.opendev.org/c/openstack/kolla-ansible/+/896292 to allow injecting
kubeconfig
Hello. Would we add helm in container as default?
mnaser
commented
8 months ago I think you're right, we are missing helm in there.
bartekle
commented
5 months ago Regarding proxy service (nice work there btw.), because I assume this also should be considered here in case of kolla-ansible type installation. Do you have some documentation about how this should be done?
Since we're more familiar with OSA, we can probably do #152 first and build this off of the same.