Fasten Ansible build progress

[YuxiangZhuang]

Fasten Ansible build using ansible.builtin.shell instead of ansible.builtin.file in node-template.yml, and add .j2 prefix of all template file.

With test, the time cost of ansible-playbook build.yaml reduce from about 59s to about 22s

[pan3793]

Encounter failures during manual verification. (Ubuntu 24.04, x64, Python 3.11, Ansible 2.15)

$ ansible-playbook build.yaml
...
TASK [orchestrate hadoop-master1 - clean and copy templates files to hadoop-master1] **************************************************************************************************************************************************************************************************************
fatal: [local]: FAILED! => {"changed": false, "msg": "Unsupported parameters for (ansible.legacy.command) module: warn. Supported parameters include: _raw_params, _uses_shell, argv, chdir, creates, executable, removes, stdin, stdin_add_newline, strip_empty_ends."}

PLAY RECAP ****************************************************************************************************************************************************************************************************************************************************************************************
local                      : ok=5    changed=0    unreachable=0    failed=1    skipped=0    rescued=0    ignored=0

ansible-playbook build.yaml  2.07s user 0.34s system 99% cpu 2.427 total

(hadoop-testing) ➜  hadoop-testing git:(PR_118) ansible --version
ansible [core 2.15.13]
  config file = /home/chengpan/Projects/hadoop-testing/ansible.cfg
  configured module search path = ['/home/chengpan/.ansible/plugins/modules', '/usr/share/ansible/plugins/modules']
  ansible python module location = /home/chengpan/.pyenv/versions/3.11.11/envs/hadoop-testing/lib/python3.11/site-packages/ansible
  ansible collection location = /home/chengpan/.ansible/collections:/usr/share/ansible/collections
  executable location = /home/chengpan/.pyenv/versions/hadoop-testing/bin/ansible
  python version = 3.11.11 (main, Jan 24 2025, 02:22:26) [GCC 13.3.0] (/home/chengpan/.pyenv/versions/3.11.11/envs/hadoop-testing/bin/python)
  jinja version = 3.1.5
  libyaml = True

[pan3793]

I don't get the idea of copying j2 templates to the destination first and deleting them eventually, why is it faster than in-place rendering?

[pan3793]

this is definitely the most costly step, in my experiment, it costs 15.94s, so collecting and dedup the dirs first and only creating the unique dirs should be a good idea