There are two things going on here. The first is that you have created YAML that is accepted by the parser, but behaves in a slightly unexpected way (and will produce a warning in the current version of Ansible.)
- name: Create a json object to use as tag
set_fact:
tag_var: '{ "test": "thisisatest" }'
set_fact:
tag_var_json: "{{ tag_var | to_json }}"
Keys in YAML are unique; when the parser encounters a second instance of the same key it throws away the first one. Since you've repeated set_fact, this is equivalent to:
- name: Create a json object to use as tag
set_fact:
tag_var_json: "{{ tag_var | to_json }}"
Correcting the syntax error will still result in a failure, however.
- name: Create a json object to use as tag
set_fact:
tag_var: '{ "test": "thisisatest" }'
tag_var_json: "{{ tag_var | to_json }}"
The arguments to set_fact have to be templated before the task runs, at which point tag_var is still undefined (because this task is defining it.)
One correct way to write this task is as two separate tasks:
- name: Create a tag object
set_fact:
tag_var:
test: thisisatest
- name: Create a JSON string for tagging
set_fact:
tag_var_json: "{{ tag_var | to_json }}"
However, set_fact isn't required at all. You can just set the var directly on the task where you use it, which is both more efficient and makes it more tightly scoped.
- name: Update instance tags
command: oci compute instance update -c {{ compartment }} --freeform-tags "{{ tag_var | to_json }}"
vars:
tag_var:
test: thisisatest
