Ansible not executing rds_instance task due to wrong variable evaluation - amazon-web-services

Question

I am struggeling with ansible since a task defined a playbook is not being executed to manage an AWS rds instance.
This is the command I execute within an jenkins pipeline:
state: "running"
identifier: "myDatabase"
sh "ansible-playbook ${env.WORKSPACE}/cost-optimization/ansible/manage_rds.yml --extra-vars 'instanceState=${state} identifier=${dbsIdentifier}'"
The playbook looks like this:
manage_rds.yml:
---
- hosts: localhost
vars:
rdsState: "{{instanceState}}"
rdsIdentifier: "{{identifier|lower}}"
tasks:
- name: "Starting RDS instances"
rds_instance:
state: running
db_instance_identifier: "{{ rdsIdentifier }}"
wait: yes
register: rds_result
when: rdsState == "running"
- name: "Stopping RDS instances"
rds_instance:
state: stopping
db_instance_identifier: "{{ rdsIdentifier }}"
wait: yes
register: rds_result
when: rdsState == "stopped"
- name: Show RDS result
debug:
var: rds_result
- import_tasks: tasks/task_create_partial_report.yml
vars:
identifier: "{{rdsIdentifier|lower}}"
partial: "db"
I expect the AWS RDS instance is being spinned up.
Instead the result looks like this:
TASK [Starting RDS instances] **************************************************
ok: [localhost]
TASK [Stopping RDS instances] **************************************************
skipping: [localhost]
TASK [Show RDS result] *********************************************************
ok: [localhost] => {
"rds_result": {
"changed": false,
"skip_reason": "Conditional result was False",
"skipped": true
}
}
Any idea how to solve this?
EDIT:
I followed the recommendation from below.
However, the RDS instance is still not affected:
+ ansible-playbook /var/lib/jenkins/workspace/.../ansible/manage_rds.yml --extra-vars 'instanceState=running identifier=myDatabase'
PLAY [localhost] ***************************************************************
TASK [Gathering Facts] *********************************************************
ok: [localhost]
TASK [Starting RDS instances] **************************************************
ok: [localhost]
TASK [Show RDS result] *********************************************************
ok: [localhost] => {
"rds_result": {
"allocated_storage": 20,
"associated_roles": [],
"auto_minor_version_upgrade": false,
"availability_zone": "eu-central-1a",
"backup_retention_period": 21,
"ca_certificate_identifier": "rds-ca-2015",
"changed": false,
"character_set_name": "SQL_Latin1_General_CP1_CI_AS",
"copy_tags_to_snapshot": true,
"db_instance_arn": "arn:aws:rds:mydatabase",
"db_instance_class": "db.t2.micro",
"db_instance_identifier": "mydatabase",
"db_instance_port": 0,
"db_instance_status": "stopped",
"db_parameter_groups": [
{
"db_parameter_group_name": "....-sqlserver-ex-14-00",
"parameter_apply_status": "in-sync"
}
],
...
So the state remains stopped althoug it is set to running.

Answer 1

register will register the result of a task, what ever that is even, if it is skipped.
In your case, you run the first task when: rdsState == "running", register the result of that successful run and immediately override your registered var with a skipped task result.
There are several solutions to go other this. I'll just give 2 out of my hat here:
Debug the var after each task
You could add the following task after each rds_instance call
- name: Show RDS result
debug:
var: rds_result
when: rds_result is not skipped
That would only print out after each task if it was actually run.
Run a single parameterized task
You need a mapping between your variable value and the one expected in the module params. Add the following to your vars:
my_rds_state:
running:
name: started
description: Starting
stopped:
name: stopped
description: Stopping
Then you can have a single task that will do the job whatever the param:
- name: "{{ my_rds_state[rdsState].description }} RDS instances"
rds_instance:
state: "{{ my_rds_state[rdsState].name }}"
db_instance_identifier: "{{ rdsIdentifier }}"
wait: yes
register: rds_result
- name: Show RDS result
debug:
var: rds_result