Ansible changes variable values when set - string

Question

I have a series of variables set. Call them Hosts and Inthosts. Each has an appropriate value set, as seen in the debug output. When I try to assign the value of inthosts to hosts, it does not actually make them the same, it bolloxes it up changing the double quotes to single quotes and putting a "u" in front of each "item". It there a way to force Ansible to actually do a literal equals in this case without parsing the text? The text should just be treated as a string. In this case the "modified" value is being output to a file, and the change breaks things.
The plan was to use the default hosts, and override it with inthosts if the server in question should be using a different set of servers.
Default Variables Set
filebeat_kafka_hosts: '["x.compute-1.amazonaws.com:9093", "y.compute-1.amazonaws.com:9093"]'
filebeat_kafka_inthosts: '["x.compute-1.amazonaws.com:9093", "y.compute-1.amazonaws.com:9093", "z.compute-1.amazonaws.com:9093"]'
Ansible Code
- debug:
msg: "Hosts {{ filebeat_kafka_hosts }} "
- debug:
msg: "IntHosts {{ filebeat_kafka_inthosts }} "
- set_fact:
filebeat_kafka_hosts="{{ filebeat_kafka_inthosts }}"
- debug:
msg: "Inthosts -> hosts {{ filebeat_kafka_hosts }} "
Output (edited)
"msg": "Hosts [\"x.compute-1.amazonaws.com:9093\", \"y.compute-1.amazonaws.com:9093\"] " |
"msg": "IntHosts [\"x.compute-1.amazonaws.com:9093\", \"y.compute-1.amazonaws.com:9093\", \"z.compute-1.amazonaws.com:9093\"] "
set {"ansible_facts": {"filebeat_kafka_hosts": ["x.compute-1.amazonaws.com:9093", "y.compute-1.amazonaws.com:9093", "z.compute-1.amazonaws.com:9093"]}, "changed": false}
"msg": "Inthosts -> hosts [u'x.compute-1.amazonaws.com:9093', u'y.compute-1.amazonaws.com:9093', u'z.compute-1.amazonaws.com:9093'] "

Answer 1

Ansible is interpreting filebeat_kafka_inthosts and filebeat_kafka_hosts as lists. That gives you the 'u' characters before each item in your debug. The tasks below
- debug:
msg: "{{ item }}"
with_items: "{{ filebeat_kafka_hosts }}"
- debug:
msg: "{{ item }}"
with_items: "{{ filebeat_kafka_inthosts }}"
Would give you
TASK [debug] *******************************************************************
ok: [127.0.0.1] => (item=y.compute-1.amazonaws.com:9093) => {
"item": "y.compute-1.amazonaws.com:9093",
"msg": "y.compute-1.amazonaws.com:9093"
}
ok: [127.0.0.1] => (item=x.compute-1.amazonaws.com:9093) => {
"item": "x.compute-1.amazonaws.com:9093",
"msg": "x.compute-1.amazonaws.com:9093"
}
TASK [debug] *******************************************************************
ok: [127.0.0.1] => (item=x.compute-1.amazonaws.com:9093) => {
"item": "x.compute-1.amazonaws.com:9093",
"msg": "x.compute-1.amazonaws.com:9093"
}
ok: [127.0.0.1] => (item=y.compute-1.amazonaws.com:9093) => {
"item": "y.compute-1.amazonaws.com:9093",
"msg": "y.compute-1.amazonaws.com:9093"
}
ok: [127.0.0.1] => (item=z.compute-1.amazonaws.com:9093) => {
"item": "z.compute-1.amazonaws.com:9093",
"msg": "z.compute-1.amazonaws.com:9093"
}
Since your writing this line to a file, you shouldn't have to worry about it too much. The 'u' character is a side effect of the debug module. Writing the variable to a file would give the same result (although with single quotes instead of double).
- lineinfile:
path: some_file
line: "{{ filebeat_kafka_hosts }}"
some_file
['x.compute-1.amazonaws.com:9093', 'y.compute-1.amazonaws.com:9093', 'z.compute-1.amazonaws.com:9093']
If you really need double quotes, you can use the to_json filter
- lineinfile:
path: some_file
line: "{{ filebeat_kafka_hosts | to_json }}"
some_file
["x.compute-1.amazonaws.com:9093", "y.compute-1.amazonaws.com:9093", "z.compute-1.amazonaws.com:9093"]

Answer 2

Seems that casting the variable as a string also works. Thanks for the advice!