Bug 1590597

Description of problem:

OSP13 minor update: common_deploy_steps_tasks.yaml runs twice for each role which results in a long minor update duration caused by redundant task.

Let's take as an example updating a simple Networker role which runs just a few services:

openstack overcloud update run --nodes Networker --playbook all

we can see this results in the following ansible-playbook commands being run:

First one:

[root@undercloud-0 stack]# cat /tmp/ansible-mistral-action3UP8t2/ansible-playbook-command.sh 
#!/bin/bash

HOME="/tmp/ansible-mistral-action3UP8t2"
ANSIBLE_LOG_PATH="/var/log/mistral/package_update.log"
ANSIBLE_HOST_KEY_CHECKING="False"
ANSIBLE_CONFIG="/tmp/ansible-mistral-action3UP8t2/ansible.cfg"

ansible-playbook -v /var/lib/mistral/204b7661-006f-4398-908f-076939b895ca/update_steps_playbook.yaml --limit Networker --module-path /usr/share/ansible-modules --user heat-admin --become --become-user root --inventory-file /tmp/ansible-mistral-action3UP8t2/inventory.yaml --private-key /tmp/ansible-mistral-action3UP8t2/ssh_private_key "$@"


and 

Second one:

[root@undercloud-0 stack]# cat /tmp/ansible-mistral-actionwy3dr5/ansible-playbook-command.sh 

#!/bin/bash

HOME="/tmp/ansible-mistral-actionwy3dr5"
ANSIBLE_LOG_PATH="/var/log/mistral/package_update.log"
ANSIBLE_HOST_KEY_CHECKING="False"
ANSIBLE_CONFIG="/tmp/ansible-mistral-actionwy3dr5/ansible.cfg"

ansible-playbook -v /var/lib/mistral/3289d21e-8672-420c-8039-5d981f8925d7/deploy_steps_playbook.yaml --limit Networker --module-path /usr/share/ansible-modules --user heat-admin --become --become-user root --inventory-file /tmp/ansible-mistral-actionwy3dr5/inventory.yaml --private-key /tmp/ansible-mistral-actionwy3dr5/ssh_private_key "$@"

Checking the actual playbooks we can spot that update_steps_playbook.yaml includes both update_steps_tasks.yaml and common_deploy_steps_tasks.yaml:

 [root@undercloud-0 ~]# cat /var/lib/mistral/204b7661-006f-4398-908f-076939b895ca/update_steps_playbook.yaml
- hosts: undercloud
  name: Gather facts undercloud
  gather_facts: yes
  become: false
- hosts: overcloud
  name: Gather facts overcloud
  gather_facts: yes
- hosts: all
  name: Load global variables
  gather_facts: no
  tasks:
    - include_vars: global_vars.yaml
- hosts: overcloud
  name: Run update
  serial: 1
  gather_facts: no
  tasks:
    - include: update_steps_tasks.yaml
      with_sequence: start=0 end=5
      loop_control:
        loop_var: step
    - include: common_deploy_steps_tasks.yaml
      with_sequence: start=1 end=5
      loop_control:
        loop_var: step


Now checking the contents of deploy_steps_playbook.yaml we can see that it also includes common_deploy_steps_tasks.yaml so these are duplicating the ones in update_steps_playbook.yaml:


 [root@undercloud-0 ~]# cat /var/lib/mistral/3289d21e-8672-420c-8039-5d981f8925d7/deploy_steps_playbook.yaml
- hosts: undercloud
  name: Gather facts undercloud
  gather_facts: yes
  become: false
  tags:
    - facts
- hosts: overcloud
  name: Gather facts overcloud
  gather_facts: yes
  tags:
    - facts
- hosts: all
  name: Load global variables
  gather_facts: no
  tasks:
    - include_vars: global_vars.yaml
  tags:
    - always
- hosts: overcloud
  name: Bootstrap TripleO servers
  gather_facts: no
  any_errors_fatal: yes
  roles:
    - tripleo-bootstrap
  tags:
    - bootstrap
- hosts: overcloud
  name: Server deployments
  gather_facts: no
  any_errors_fatal: yes
  tasks:
    - include: ControllerOpenstack/deployments.yaml
      vars:
        force: false
      when: role_name == 'ControllerOpenstack'
      with_items: "{{ ControllerOpenstack_pre_deployments|default([]) }}"
    - include: Database/deployments.yaml
      vars:
        force: false
      when: role_name == 'Database'
      with_items: "{{ Database_pre_deployments|default([]) }}"
    - include: Messaging/deployments.yaml
      vars:
        force: false
      when: role_name == 'Messaging'
      with_items: "{{ Messaging_pre_deployments|default([]) }}"
    - include: Networker/deployments.yaml
      vars:
        force: false
      when: role_name == 'Networker'
      with_items: "{{ Networker_pre_deployments|default([]) }}"
    - include: CephStorage/deployments.yaml
      vars:
        force: false
      when: role_name == 'CephStorage'
      with_items: "{{ CephStorage_pre_deployments|default([]) }}"
    - include: Compute/deployments.yaml
      vars:
        force: false
      when: role_name == 'Compute'
      with_items: "{{ Compute_pre_deployments|default([]) }}"
  tags:
    - overcloud
    - pre_deploy_steps
- hosts: overcloud
  name: Host prep steps
  gather_facts: no
  any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
  become: true
  tasks:
    - include: ControllerOpenstack/host_prep_tasks.yaml
      when: role_name == 'ControllerOpenstack'
    - include: Database/host_prep_tasks.yaml
      when: role_name == 'Database'
    - include: Messaging/host_prep_tasks.yaml
      when: role_name == 'Messaging'
    - include: Networker/host_prep_tasks.yaml
      when: role_name == 'Networker'
    - include: CephStorage/host_prep_tasks.yaml
      when: role_name == 'CephStorage'
    - include: Compute/host_prep_tasks.yaml
      when: role_name == 'Compute'
  tags:
    - overcloud
    - host_prep_steps
- hosts: undercloud
  name: External deployment step 1
  gather_facts: no
  any_errors_fatal: yes
  become: false
  vars:
    step: '1'
  tasks:
    - include: external_deploy_steps_tasks.yaml
  tags:
    - external
    - external_deploy_steps
- hosts: overcloud
  name: Overcloud deploy step tasks for 1
  gather_facts: no
  any_errors_fatal: yes
  # FIXME(shardy) - it would be nice to use strategy: free to
  # allow the tasks per-step to run in parallel on each role,
  # but that doesn't work with any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '1'
  tasks:
    - include: ControllerOpenstack/deploy_steps_tasks.yaml
      when: role_name == 'ControllerOpenstack'
    - include: Database/deploy_steps_tasks.yaml
      when: role_name == 'Database'
    - include: Messaging/deploy_steps_tasks.yaml
      when: role_name == 'Messaging'
    - include: Networker/deploy_steps_tasks.yaml
      when: role_name == 'Networker'
    - include: CephStorage/deploy_steps_tasks.yaml
      when: role_name == 'CephStorage'
    - include: Compute/deploy_steps_tasks.yaml
      when: role_name == 'Compute'
  tags:
    - overcloud
    - deploy_steps
- hosts: overcloud
  name: Overcloud common deploy step tasks 1
  gather_facts: no
  any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '1'
  tasks:
    - include: common_deploy_steps_tasks.yaml
  tags:
    - overcloud
    - deploy_steps
- hosts: undercloud
  name: External deployment step 2
  gather_facts: no
  any_errors_fatal: yes
  become: false
  vars:
    step: '2'
  tasks:
    - include: external_deploy_steps_tasks.yaml
  tags:
    - external
    - external_deploy_steps
- hosts: overcloud
  name: Overcloud deploy step tasks for 2
  gather_facts: no
  any_errors_fatal: yes
  # FIXME(shardy) - it would be nice to use strategy: free to
  # allow the tasks per-step to run in parallel on each role,
  # but that doesn't work with any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '2'
  tasks:
    - include: ControllerOpenstack/deploy_steps_tasks.yaml
      when: role_name == 'ControllerOpenstack'
    - include: Database/deploy_steps_tasks.yaml
      when: role_name == 'Database'
    - include: Messaging/deploy_steps_tasks.yaml
      when: role_name == 'Messaging'
    - include: Networker/deploy_steps_tasks.yaml
      when: role_name == 'Networker'
    - include: CephStorage/deploy_steps_tasks.yaml
      when: role_name == 'CephStorage'
    - include: Compute/deploy_steps_tasks.yaml
      when: role_name == 'Compute'
  tags:
    - overcloud
    - deploy_steps
- hosts: overcloud
  name: Overcloud common deploy step tasks 2
  gather_facts: no
  any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '2'
  tasks:
    - include: common_deploy_steps_tasks.yaml
  tags:
    - overcloud
    - deploy_steps
- hosts: undercloud
  name: External deployment step 3
  gather_facts: no
  any_errors_fatal: yes
  become: false
  vars:
    step: '3'
  tasks:
    - include: external_deploy_steps_tasks.yaml
  tags:
    - external
    - external_deploy_steps
- hosts: overcloud
  name: Overcloud deploy step tasks for 3
  gather_facts: no
  any_errors_fatal: yes
  # FIXME(shardy) - it would be nice to use strategy: free to
  # allow the tasks per-step to run in parallel on each role,
  # but that doesn't work with any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '3'
  tasks:
    - include: ControllerOpenstack/deploy_steps_tasks.yaml
      when: role_name == 'ControllerOpenstack'
    - include: Database/deploy_steps_tasks.yaml
      when: role_name == 'Database'
    - include: Messaging/deploy_steps_tasks.yaml
      when: role_name == 'Messaging'
    - include: Networker/deploy_steps_tasks.yaml
      when: role_name == 'Networker'
    - include: CephStorage/deploy_steps_tasks.yaml
      when: role_name == 'CephStorage'
    - include: Compute/deploy_steps_tasks.yaml
      when: role_name == 'Compute'
  tags:
    - overcloud
    - deploy_steps
- hosts: overcloud
  name: Overcloud common deploy step tasks 3
  gather_facts: no
  any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '3'
  tasks:
    - include: common_deploy_steps_tasks.yaml
  tags:
    - overcloud
    - deploy_steps
- hosts: undercloud
  name: External deployment step 4
  gather_facts: no
  any_errors_fatal: yes
  become: false
  vars:
    step: '4'
  tasks:
    - include: external_deploy_steps_tasks.yaml
  tags:
    - external
    - external_deploy_steps
- hosts: overcloud
  name: Overcloud deploy step tasks for 4
  gather_facts: no
  any_errors_fatal: yes
  # FIXME(shardy) - it would be nice to use strategy: free to
  # allow the tasks per-step to run in parallel on each role,
  # but that doesn't work with any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '4'
  tasks:
    - include: ControllerOpenstack/deploy_steps_tasks.yaml
      when: role_name == 'ControllerOpenstack'
    - include: Database/deploy_steps_tasks.yaml
      when: role_name == 'Database'
    - include: Messaging/deploy_steps_tasks.yaml
      when: role_name == 'Messaging'
    - include: Networker/deploy_steps_tasks.yaml
      when: role_name == 'Networker'
    - include: CephStorage/deploy_steps_tasks.yaml
      when: role_name == 'CephStorage'
    - include: Compute/deploy_steps_tasks.yaml
      when: role_name == 'Compute'
  tags:
    - overcloud
    - deploy_steps
- hosts: overcloud
  name: Overcloud common deploy step tasks 4
  gather_facts: no
  any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '4'
  tasks:
    - include: common_deploy_steps_tasks.yaml
  tags:
    - overcloud
    - deploy_steps
- hosts: undercloud
  name: External deployment step 5
  gather_facts: no
  any_errors_fatal: yes
  become: false
  vars:
    step: '5'
  tasks:
    - include: external_deploy_steps_tasks.yaml
  tags:
    - external
    - external_deploy_steps
- hosts: overcloud
  name: Overcloud deploy step tasks for 5
  gather_facts: no
  any_errors_fatal: yes
  # FIXME(shardy) - it would be nice to use strategy: free to
  # allow the tasks per-step to run in parallel on each role,
  # but that doesn't work with any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '5'
  tasks:
    - include: ControllerOpenstack/deploy_steps_tasks.yaml
      when: role_name == 'ControllerOpenstack'
    - include: Database/deploy_steps_tasks.yaml
      when: role_name == 'Database'
    - include: Messaging/deploy_steps_tasks.yaml
      when: role_name == 'Messaging'
    - include: Networker/deploy_steps_tasks.yaml
      when: role_name == 'Networker'
    - include: CephStorage/deploy_steps_tasks.yaml
      when: role_name == 'CephStorage'
    - include: Compute/deploy_steps_tasks.yaml
      when: role_name == 'Compute'
  tags:
    - overcloud
    - deploy_steps
- hosts: overcloud
  name: Overcloud common deploy step tasks 5
  gather_facts: no
  any_errors_fatal: yes
  vars:
    bootstrap_server_id: 2001f017-6bb9-435b-8b46-ceb2da0908a3
    step: '5'
  tasks:
    - include: common_deploy_steps_tasks.yaml
  tags:
    - overcloud
    - deploy_steps
- hosts: overcloud
  name: Server Post Deployments
  gather_facts: no
  any_errors_fatal: yes
  tasks:
    - include: ControllerOpenstack/deployments.yaml
      vars:
        force: false
      when: role_name == 'ControllerOpenstack'
      with_items: "{{ ControllerOpenstack_post_deployments|default([]) }}"
    - include: Database/deployments.yaml
      vars:
        force: false
      when: role_name == 'Database'
      with_items: "{{ Database_post_deployments|default([]) }}"
    - include: Messaging/deployments.yaml
      vars:
        force: false
      when: role_name == 'Messaging'
      with_items: "{{ Messaging_post_deployments|default([]) }}"
    - include: Networker/deployments.yaml
      vars:
        force: false
      when: role_name == 'Networker'
      with_items: "{{ Networker_post_deployments|default([]) }}"
    - include: CephStorage/deployments.yaml
      vars:
        force: false
      when: role_name == 'CephStorage'
      with_items: "{{ CephStorage_post_deployments|default([]) }}"
    - include: Compute/deployments.yaml
      vars:
        force: false
      when: role_name == 'Compute'
      with_items: "{{ Compute_post_deployments|default([]) }}"
  tags:
    - overcloud
    - post_deploy_steps
- hosts: undercloud
  name: External deployment Post Deploy tasks
  gather_facts: no
  any_errors_fatal: yes
  become: false
  tasks:
    - include: external_post_deploy_steps_tasks.yaml
  tags:
    - external
    - external_deploy_steps


Version-Release number of selected component (if applicable):
openstack-tripleo-heat-templates-8.0.2-33.el7ost.noarch
openstack-tripleo-common-8.6.1-20.el7ost.noarch
python-tripleoclient-9.2.1-12.el7ost.noarch

How reproducible:
100%

Steps to Reproduce:

1. Deploy OSP13
2. Run minor update procedure
3. Check the resulting ansible playbook

Actual results:
Duplicated common_deploy_steps_tasks.yaml are included.

Expected results:
No unnecessary tasks are run.

Additional info:

This results in long minor update times, to give an example, to update *only* 3 controller nodes it takes 1h and 30minutes. 


To update 2 networker nodes it takes 23 minutes. If I try to re-run the update on the networker nodes after minor update(basically a noop since everything is up to date) it takes 13 minutes which is a lot imo just for a noop for 2 simple nodes.