Skip to content

Latest commit

 

History

History
142 lines (122 loc) · 6.33 KB

README.md

File metadata and controls

142 lines (122 loc) · 6.33 KB

Single node openshift installation

This is a set of playbooks to automate the deployment of Openshift cluster with single node footprint, using Assisted Installer and Small ISO.

Requirements

  1. Podman.
  2. Ansible.
  3. Ansible modules containers.podman and community.general. You can install all the requirements with:
ansible-galaxy collection install -r requirements.yaml

Steps

  1. git clone the ztp-cluster-deploy repo

  2. go to ai-deploy-cluster-singlenode dir

  3. You need make a copy inventory/hosts.sample file and name it hosts under same directory.

  4. Modify the [all:vars] section at inventory/hosts file based on your HW env.

  5. Start the deployment with prepare-environment tag:

    $ sudo ansible-playbook playbook.yml -vvv -i inventory/hosts --tags=prepare-environment

  6. Make sure that AI pod has been created and the http server is running:

    $ sudo systemctl status httpd
httpd.service - The Apache HTTP Server
  Loaded: loaded (/usr/lib/systemd/system/httpd.service; enabled; vendor preset: disabled)
  Active: active (running) since Mon 2020-12-14 14:21:12 EST; 5h 55min ago
    $ sudo podman ps
CONTAINER ID  IMAGE                                          COMMAND               CREATED      STATUS          PORTS                   NAMES
2a9f6e4056a8  k8s.gcr.io/pause:3.2                                                 6 hours ago  Up 6 hours ago  0.0.0.0:5432->5432/tcp  2ae5adf6da27-infra
603d56566a6b  quay.io/ocpmetal/postgresql-12-centos7:latest  run-postgresql        6 hours ago  Up 6 hours ago  0.0.0.0:5432->5432/tcp  db
a4d2a578fe32  quay.io/ocpmetal/assisted-service:latest       /assisted-service     6 hours ago  Up 6 hours ago  0.0.0.0:5432->5432/tcp  installer
f8ab3586816f  quay.io/ocpmetal/ocp-metal-ui:latest           /opt/bitnami/scri...  6 hours ago  Up 6 hours ago  0.0.0.0:5432->5432/tcp  ui

    Note: Apache httpd is installed by the playbook and used to serve the generated iso files during installation. You can use any other HTTP server, but you will need to adjust configurations in the inventory/hosts file. For httpd, set as follows:

    final_iso_path=/var/www/html

    Note: Configure or disable your firewall to avoid any tcp ports blocking issues

  7. Based on your env and your machine intf use nmcli command to set the network-config under /opt dir

    $ sudo mkdir /opt/network-config
Note: Then you can config your machine as below
$ cat /opt/network-config/etc/NetworkManager/system-connections/eno1.nmconnection 
[connection]
id=static
type=ethernet
interface-name=eno1
permissions=

[ethernet]
mac-address-blacklist=

[ipv4]
method=auto
dhcp-hostname=cluster.local
dhcp-send-hostname=true

[ipv6]
method=ignore

  8. Finally, you can execute the rest of installation steps as shown below:

    $ sudo ansible-playbook playbook.yml -vvv -i inventory/hosts --skip-tags=prepare-environment

    Note: Based on your env the installation should take around 1h to 1:30h.

    For any error that appears, you can re-run any installation step by its tag name (just specify the tag as we did in step 5).

    $ cat playbook.yml 
  ---
  - name: Deploy single node cluster with assisted installer
    hosts: provisioner
    environment:
      PATH: "/usr/bin/:/usr/local/bin/:{{ ansible_env.PATH }}"
    roles:
      - role: prepare-environment
        tags: prepare-environment
        become: yes
      - role: create-cluster
        tags: create-cluster
        become: yes
      - role: add-baremetal-node
        tags: add-baremetal-node
        become: yes
      - role: deploy-cluster
        tags: deploy-cluster
        become: yes

After deployment

In the provisioner, use the assisted installer commands to check the created clusters and nodes. The installation goes through phases of installing and finalizing to installed.

The assisted installer CLI (aicli) must have the AI endpoint defined as in your inventory/hosts. You can specify it as follows:

$ export AICLI_URL=<ai_url from your inventory>

Alternatively, it can be added to the aicli command with -U key.

After the successful case you should see an output similar to the below:

$ aicli list cluster
+-------------+--------------------------------------+--------------+----------------------------+
|   Cluster   |                  Id                  |    Status    |         Dns Domain         |
+-------------+--------------------------------------+--------------+----------------------------+
|   test-aut  | eaf6b793-d2b0-496d-827a-93abd8e041af |  installed   |      cluster.testing       |
+-------------+--------------------------------------+--------------+----------------------------+
$ aicli list hosts
+----------------------------------------------+----------+--------------------------------------+-----------+--------+----------------+
|                     Host                     | Cluster  |                  Id                  |   Status  |  Role  |       Ip       |
+----------------------------------------------+----------+--------------------------------------+-----------+--------+----------------+
| single-node-cluster.test-aut.cluster.testing | test-aut | 0471f8d5-812a-1e08-42e0-27e173750188 | installed | master | 192.168.112.34 |
+----------------------------------------------+----------+--------------------------------------+-----------+--------+----------------+

Now you can download the kubeconfig of your cluster using the command below:

$ aicli download kubeconfig {YOUR_CLUSTER_NAME} --path ~/

Use the downloaded kubeconfig file with OC command to control the openshift cluster. You should have one node with master/worker roles:

$ export KUBECONFIG=<path to your kubeconfig file>
$ oc get nodes
NAME                                           STATUS   ROLES           AGE   VERSION
single-node-cluster.test-aut.cluster.testing   Ready    master,worker   22m   v1.20.0+d9c52cc