Skip to content

Latest commit

 

History

History
 
 

ocp-workloads

readme.adoc

Deploy OpenShift workloads to a cluster or clusters

Table of Contents

This config is used to run OpenShift workloads on a single cluster or a list of clusters.

It is a config like any other agnosticd config and allows integration of this particular kind of deployment into our eco-system.

It can access a single cluster via a bastion or multiple clusters by direct connection the the cluster apis.

Create

Example: using-bastion.yml
cloud_provider: none
env_type: ocp-workloads

ocp_workloads:
  - ocp4-workload-fuse-ignite

target_host: bastion.dev.mycluster.mydomain.com

#become_override: false

# If the ocp-workload supports it, you should specify the OCP user:
# ocp_username: myuser

# Usually the ocp-workloads want a GUID also:
# guid: changeme
Example: multicluster-bastion.yml
cloud_provider: none
guid: changeme
env_type: ocp-workloads

cluster_workloads:
- name: ocp4_workload_authentication
  clusters:
  - openshift_app
  - openshift_db
- name: ocp4_workload_mitzi_app
  clusters:
  - openshift_app
- name: ocp4_workload_mitzi_db
  clusters:
  - openshift_db

openshift_app:
  api_ca_cert: ...
  api_key: ...
  api_url: ...
  ocp4_workload_authentication_htpasswd_user_name: app-user

openshift_db:
  api_ca_cert: ...
  api_key: ...
  api_url: ...
  ocp4_workload_authentication_htpasswd_user_name: db-user

ocp4_workload_authentication_idm_type: htpasswd
ocp4_workload_authentication_htpasswd_user_count: 1
Run
cd agnosticd/ansible
ansible-playbook main.yml -e @vars.yml

Delete

Just run the following:

cd agnosticd/ansible
ansible-playbook destroy.yml -e @vars.yml

It will run the ocp-workload role with ACTION=destroy.

Lifecycle

The lifecycle_entry_point.yml playbook can be used as well.

It will just run the workload passing the ACTION variable. Just make sure to implement the action stop, start, status in the ocp-workloads.

cluster_workloads variable

This variable allows configuration to run workloads across multiple clusters. This feature is currently only supported with direct connection from localhost.

The cluster_workloads array must specify target clusters whose names correspond to variables than hold connection configuration for the cluster as shown in the example above.

Cluster specific workload variables may be passed within the cluster variables. Variables must follow the convention of beginning with the workload name followed by an underscore and not be set at the top-level as the set_fact call used to pass the variable to the workload cannot override a variable passed with Ansible extra vars.

target_host variable

This variable correspond to run workloads on a single cluster using a bastion host to run oc commands or k8s type ansible tasks.

You can specify the target host in three different ways.

As localhost

target_host: localhost

This will execute the workload from localhost. It requires that whatever host the playbook is run from is authenticated to the OpenShift cluster.

As an hostname

If you want to just specify:

target_host: bastion.dev.mycluster.mydomain.com

Then you need to configure ssh properly to be able to connect to that host. Just make sure the command ssh bastion.dev.mycluster.mydomain.com works.

As a dictionary

You can specify the bastion host using a dictionary. This is useful is you need to specify the user, port, ssh_key to use, etc.

target_host:
  ansible_host: bastion.babydev.babylon.open.redhat.com
  ansible_port: 22
  ansible_user: ec2-user
  #ansible_ssh_private_key_content: "{{ ssh_private_key_content }}"
  ansible_ssh_private_key_file: ~/.ssh/admin_key.pem
  #ansible_ssh_extra_args:  ...
Note
 you can add the ansible_ssh_private_key_content to a secret file or a vault. The config will create the key using that content in the directory output_dir/ and use it to connect to the bastion. The key will then be deleted when the playbook ends, see cleanup.yml.

FAQ

  1. But i want to run my workload as root on the bastion!

Just use the var become_override. Set it to true in your var file. Most ocp-workloads implement that variable.

extract of main.yml in ocp-workload
- name: Running Workload Tasks
  import_tasks: ./workload.yml
  become: "{{ become_override | bool }}"
  when: ACTION == "create" or ACTION == "provision"