Select one usable server from inventory - ansible

Question

I have a group of servers, which may or may not be available at the moment. I want to execute queries against a usable server. I don't know which one to use, though. So I thought I could use an include...until construct, but that doesn't work as expected.
Example:
main.yml
---
- hosts: localhost
tasks:
- include: get_available_node.yml
loop:
- localhost
- example.com
- localhost
loop_control:
loop_var: domain
until: available_domain is defined and available_domain|length > 0
- debug:
var: available_domain
get_available_node.yml
---
- get_url:
url: "http://{{ domain }}"
dest: /dev/null
register: url_attempt
ignore_errors: yes
- set_fact:
available_domain: "{{ domain }}"
when: url_attempt.failed == false
I thought, my loop should stop as soon as my available_domain var gets set. However, it doesn't.
PLAY [localhost] ******************************************************************************************************************************************************************
TASK [Gathering Facts] ************************************************************************************************************************************************************
ok: [localhost]
TASK [include] ********************************************************************************************************************************************************************
included: /Volumes/Source Code/playground/stackoverflow/get_available_node.yml for localhost => (item=localhost)
included: /Volumes/Source Code/playground/stackoverflow/get_available_node.yml for localhost => (item=example.com)
included: /Volumes/Source Code/playground/stackoverflow/get_available_node.yml for localhost => (item=localhost)
TASK [Try to curl localhost] ******************************************************************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": false, "dest": "/dev/null", "elapsed": 0, "gid": 0, "group": "wheel", "mode": "0666", "msg": "Request failed: <urlopen error [Errno 61] Connection refused>", "owner": "root", "size": 0, "state": "file", "uid": 0, "url": "http://localhost"}
...ignoring
TASK [set_fact] *******************************************************************************************************************************************************************
skipping: [localhost]
TASK [Try to curl example.com] ****************************************************************************************************************************************************
ok: [localhost]
TASK [set_fact] *******************************************************************************************************************************************************************
ok: [localhost]
TASK [Try to curl localhost] ******************************************************************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": false, "dest": "/dev/null", "elapsed": 0, "gid": 0, "group": "wheel", "mode": "0666", "msg": "Request failed: <urlopen error [Errno 61] Connection refused>", "owner": "root", "size": 0, "state": "file", "uid": 0, "url": "http://localhost"}
...ignoring
TASK [set_fact] *******************************************************************************************************************************************************************
skipping: [localhost]
TASK [debug] **********************************************************************************************************************************************************************
ok: [localhost] => {
"available_domain": "example.com"
}
PLAY RECAP ************************************************************************************************************************************************************************
localhost : ok=9 changed=0 unreachable=0 failed=0 skipped=2 rescued=0 ignored=2
At the end, a usable server is in the variable, however, if there are 20 hosts in my original list, it takes a lot of time to loop through them.
Where is my misunderstanding of how the until parameter works? Is there a way to achieve what I'm trying to do?

Answer 1

Using run_once should work, as long as setup is run (ie gather_facts: true), because ansible will remove those hosts which can't be reached for Fact Gathering, from the candidate pool for run_once, and thus only run on the first "up" host. for example:
---
- hosts: all
become: false
run_once: true
gather_facts: true
tasks:
- ping:
$ ansible-playbook -i does,not,exist,localhost, test.yml
will run ping on only one of 4 hosts only: localhost, the only one that's up
PLAY RECAP ********************************************************************
does : ok=0 changed=0 unreachable=1 failed=0 skipped=0 rescued=0 ignored=0
exist : ok=0 changed=0 unreachable=1 failed=0 skipped=0 rescued=0 ignored=0
localhost : ok=2 changed=0 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
not : ok=0 changed=0 unreachable=1 failed=0 skipped=0 rescued=0 ignored=0
(you might have to add ansible_connection: local to host_vars/localhost to simulate). If there are multiple hosts up, run_once will only run on the first one from the inventory that's actually up.
Note, if you're doing fact caching, I'm not sure if this will still hold true, because setup might not get run and remove the node from candidate pool. If you need this, an initial ping task through all the servers, followed by the "work" task with run_once, might work (untested), with the default linear strategy.