I've registered a variable in a play.
---
- hosts: 127.0.0.1
gather_facts: no
connection: local
sudo: no
vars_files:
- vars.yml
tasks:
- name: build load balancer
os_load_balancer: net=mc_net ext_net=vlan3320 name=load_balancer protocol=HTTPS port=80
register: my_lb
I can access that variable fine, until I make the request inside a role.
For example, in a separate role in the same run, I want to access that registered variable:
- debug: var=my_lb
I get the following output:
{'msg': "AnsibleUndefinedVariable: One or more undefined variables: 'my_lb' is undefined", 'failed': True}
How can I access variables registered in a separate role, within the same play?
Edit for clarity of how things piece together:
Top Play
-includes:
- Sub play 1
- registers variable foo
- Sub play 2
-includes:
- sub play A
- role 1
- role 2
- role 3
- references variable foo in template
- Sub play B
- Sub play 3
NB: This was written referring to Ansible 1.X. I suspect variable parsing and scoping has changed considerably with Ansible 2.0, so bear that in mind. I'll try to update the answer if I get time (or maybe someone else will answer for v2.0!)
There are two options here. The simplest is to define the variables in your top-level playbook, then propagate them down into your various roles. You can either set these as simple vars, or use pre_tasks to lookup / calculate values dynamically, e.g.
vars:
foo_static: "value_bar"
pre_tasks:
- set_fact:
foo_ncpu: "{{ ansible_processor_vcpus }}"
roles:
- {role: role1, role1_var1: "{{foo_static}}", role1_var2: "{{foo_ncpu}}" }
- {role: role2, role2_var1: "{{foo_static}}", role2_var2: "{{foo_ncpu}}" }
The second option requires that you add a task to whichever role you need to extract a variable from (though since all ansible roles are open-source, that should be easy enough). The trick is to use set_fact to export a variable as a 'host_fact', e.g.
- name: Export role1_varfoo to a 'host-fact' type variable
set_fact:
role1_varfoo: "{{ role1_varfoo }}"
Which can then be accessed later like so:
vars:
role2_varfoo: "{{ hostvars['myhost']['role1_varfoo']}}"
As described in this bit of the ansible docs.
Note that if you always want to lookup the hostvars of the current machine you're running commands on (without knowing what it's actually called in the ansible hosts file), you can use the variable inventory_hostname like so:
vars:
role2_varfoo: "{{ hostvars[inventory_hostname]['role1_varfoo']}}"
(Note lack of quote-marks, since it's a variable, not a string-literal.)
Slightly awkward, but those combinations have met all my needs so far.
Try moving your variable declarations into a pre_task block. Variables set here should be available within and following roles.
https://docs.ansible.com/playbooks_roles.html#roles
e.g.
pre_tasks:
- name: build load balancer
os_load_balancer: net=mc_net ext_net=vlan3320 name=load_balancer protocol=HTTPS port=80
register: my_lb
roles:
- { role: some_role }
Update: To access the variable using the hostvars syntax use the appropriate host GROUP variable rather than the host that executed the set_fact:
hostvars[inventory_hostname]['variable']
To broaden the accepted question.
1.-You can define ANOTHER role where you register the variable and then set it there then refer that variable from multiple roles. AS LONG as the roles are in the same play.
Docu here:
http://docs.ansible.com/ansible/playbooks_variables.html#variable-examples
Generally speaking, variables set in one role are available to others. This means if you have a roles/common/vars/main.yml you can set variables in there and make use of them in other roles and elsewhere in your playbook
Edit: Clarification, this applies for REGISTERED and set variables in Ansible 2.x in my experience.
2.-As far as using hostvars goes, I tried it myself and failed with this error:
fatal: [localhost]: FAILED! => {"failed": true, "msg": "'dict object' has no attribute 'ec2_instance_id'"}
I my case I was doing the following.
- hosts: localhost
gather_facts: yes
roles:
- { role: role_1 }
post_tasks:
- name: Check instance variables within localhost
debug: var={{ hostvars['localhost']['ec2_instance_id'] }}
On role 1 I had:
- name: register instance_id
set_fact: ec2_instance_id="{{ item.id }}"
with_items: "{{ ec2_instance.instances }}"
Although according to this old thread, the hostvar approach should work:
https://github.com/ansible/ansible/issues/1934
How abt passing it as parameter to role. I'm assuming such (register) variables aren't passed implicitly (like hotvars are).
- { role: my_role, my_lb: "{{my_lb}}" }
Related
my playbook structure looks like:
- hosts: all
name: all
roles:
- roles1
- roles2
In tasks of roles1, I define such a variable
---
# tasks for roles1
- name: Get the zookeeper image tag # rel3.0
run_once: true
shell: echo '{{item.split(":")[-1]}}' # Here can get the string rel3.0 normally
with_items: "{{ret.stdout.split('\n')}}"
when: "'zookeeper' in item"
register: zk_tag
ret.stdout:
Loaded image: test/old/kafka:latest
Loaded image: test/new/mysql:v5.7
Loaded image: test/old/zookeeper:rel3.0
In tasks of roles2, I want to use the zk_tag variable
- name: Test if the variable zk_tag can be used in roles2
debug: var={{ zk_tag.stdout }}
Error :
The task includes an option with an undefined variable. The error was: 'dict object' has no attribute 'stdout'
I think I encountered the following 2 problems:
When registering a variable with register, when condition is added, this variable cannot be used in all groups. How to solve this problem? How to make this variable available to all groups?
is my title, How to use variables between different roles in ansible?
You're most likely starting a new playbook for a new host. Meaning all previous collected vars are lost.
What you can do is pass a var to another host with the add_host module.
- name: Pass variable from this play to the other host in the same play
add_host:
name: hostname2
var_in_play_2: "{{ var_in_play_1 }}"
--- EDIT ---
It's a bit unclear. Why do you use the when statement in the first place if you want every host in the play for it to be available?
You might want to use the group_vars/all.yml file to place vars in.
Also, using add_host should be the way to go as how I read it. Can you post your playbook, and the outcome of your playbook on a site, e.g. pastebin?
If there is any chance the var is not defined because of a when condition, you should use a default value to force the var to be defined when using it. While you are at it, use the debug module for your tests rather than echoing something in a shell
- name: Debug my var
debug:
msg: "{{ docker_exists | default(false) }}"
Given a playbook like this:
- name: "Tasks for service XYZ"
hosts: apiservers
roles:
- { role: common }
Is there a way to reference the playbook's name ("Tasks for service XYZ")? (i.e. a variable)
EDIT:
My intention is to be able to reference the playbook's name in a role task, i.e. sending a msg via slack like
- name: "Send Slack notification indicating deploy has started"
slack:
channel: '#project-deploy'
token: '{{ slack_token }}'
msg: '*Deploy started* to _{{ inventory_hostname }}_ of `{{ PLAYBOOK_NAME }}` version *{{ service_version }}*'
delegate_to: localhost
tags: deploy
It was added in 2.8:
ansible_play_name
The name of the currently executed play. Added in 2.8.
No, the special variables for Ansible are documented here, and you can see that there is no variable to return the playbook name.
As mentioned in the comments, however, you can always do this:
---
- name: "{{ task_name }}"
hosts: localhost
vars:
task_name: "Tasks for service XYZ"
tasks:
- debug:
msg: "{{ task_name }}"
From your circumstances, it looks like you only want this for audit/notification purposes? In that case (and assuming unixy clients), using
lookup('file', '/proc/self/cmdline') | regex_replace('\u0000',' ')
will give you the entire command line that ansible-playbook was called with, parameters and all, which would include the playbook name. Depending on your circumstances, that might be a useful enough tradeoff.
How can I declare global variable within Ansible playbook. I have searched in google and found the below solution, but its not working as expected.
- hosts: all
vars:
prod-servers:
- x.x.x.x
- x.x.x.x
- hosts: "{{prod-servers}}"
tasks:
- name: ping
action: ping
When I'm trying the above code, it says variable prod-servers is undefined.
You cannot define a variable accessible on a playbook level (global variable) from within a play.
Variable Scopes
Ansible has 3 main scopes:
Global: this is set by config, environment variables and the command line
Play: each play and contained structures, vars entries (vars; vars_files; vars_prompt), role defaults and vars.
Host: variables directly associated to a host, like inventory, include_vars, facts or registered task outputs
Anything you declare inside a play can thus only be either a play variable, or a (host) fact.
To define a variable, which you can use in the hosts declaration:
run ansible-playbook with --extra-vars option and pass the value in the argument;
or to achieve the same functionality (decide which hosts to run a play on, from within a preceding play):
define an in-memory inventory and run the subsequent play against that inventory.
what you seem to want is an inventory (http://docs.ansible.com/ansible/latest/intro_inventory.html), it looks like you have an static list of IP's that may be prod servers (or dev, or whatever), therefore you can create an static inventory.
In your second play you want to use the list of IP's as hosts to run the tasks, that's not what Ansible expects. After the "hosts" keyword in a play declaration, Ansible expects a group name from the inventory.
If, on the opossite, your prod servers change from time to time, you may need to create a dynamic inventory. You can have a look at examples in https://github.com/ansible/ansible/tree/devel/contrib/inventory (for instance, there are examples of dynamic inventory based on EC2 from Amazon or vsphere)
regards
well, this can be done using
set_fact.
I don't know the best practice for this but this works for me
Here's my playbook example
- hosts: all
gather_facts: false
tasks:
- set_fact: host='hostname'
- hosts: host-name1
gather_facts: false
tasks:
- name: CheckHostName
shell: "{{ host }}"
register: output
- debug: msg="{{ output }}"
- hosts: host-name2
gather_facts: false
tasks:
- name: CheckHostName
shell: "{{ host }}"
register: output
- debug: msg="{{ output }}"
Is there a variable or a method allowing one to list all the roles applied to a group of ansible hosts?
For example:
- hosts: webservers
gather_facts: true
roles:
- nginx
- php-fpm
tasks:
- debug:
msg: {{ item }} installed
with_items: ansible_roles
or perhaps another way to achieve this?
It depends on what applied means in your question.
The variable role_names holds all roles of the current play, so it would be an array: [nginx, php-fpm].
- debug:
msg: {{ item }} installed
with_items: role_names
But these roles not necessarily have been applied to the hosts, if you mean by that they have been processed. There is no such thing that will update once a role has been run on a host.
If that is what you're looking for you could implement it yourself with a callback plugin. AFAIK there is no callback for starting/completing a role itself. But since the names of the roles are present in the task names you could simply use the playbook_on_task_start, extract the role name from the task name and store it in some way. I have not yet looked into callback plugins in Ansible 2 where the API changed, but I expect you have access to the global variable list and can manipulate it.
I cannot get this seemingly simple example to work in Ansible 1.8.3. The variable interpolation does not kick in the task name. All examples I have seen seem to suggest this should work. Given that the variable is defined in the vars section I expected the task name to print the value of the variable. Why doesn't this work?
Even the example from the Ansible documentation seems to not print the variable value.
---
- hosts: 127.0.0.1
gather_facts: no
vars:
vhost: "foo"
tasks:
- name: create a virtual host file for {{ vhost }}
debug: msg="{{ vhost }}"
This results in the following output:
PLAY [127.0.0.1]
**************************************************************
TASK: [create a virtual host file for {{ vhost }}]
****************************
ok: [127.0.0.1] => {
"msg": "foo"
}
PLAY RECAP
********************************************************************
127.0.0.1 : ok=1 changed=0 unreachable=0 failed=0
Update
This works with 1.7.2 but does not work with 1.8.3. So either this is a bug or a feature.
Variables are not resolved inside the name. Only inside the actual tasks/conditions etc. the placeholders will be resolved. I guess this is by design. Imagine you have a with_items loop and use the {{ item }}in the name. The tasks name will only be printed once, but the {{ item }} would change in every iteration.
I see the examples, even the one in the doc you linked to, use variables in the name. But that doesn't mean the result would be like you expected it. The docs are community managed. It might be someone just put that line there w/o testing it - or maybe it used to work like that in a previous version of Ansible and the docs have not been updated then. (I'm only using Ansible since about one year). But even though it doesn't work like we wish it would, I'm still using variables in my name's, just to indicate that the task is based on dynamic parameters. Might be the examples have been written with the same intention.
An interesting observation I recently made (Ansible 1.9.4) is, default values are written out in the task name.
- name: create a virtual host file for {{ vhost | default("foo") }}
When executed, Ansible would show the task title as:
TASK: [create a virtual host file for foo]
This way you can avoid ugly task names in the output.
Explanation
Whether the variable gets interpolated depends on where it has been declared.
Imagine You have two hosts: A and B.
If variable foo has only per-host values, when Ansible runs the play, it cannot decide which value to use.
On the other hand, if it has a global value (global in a sense of host invariance), there is no confusion which value to use.
Source: https://github.com/ansible/ansible/issues/3103#issuecomment-18835432
Hands on playbook
ansible_user is an inventory variable
greeting is an invariant variable
- name: Test variable substitution in names
hosts: localhost
connection: local
vars:
greeting: Hello
tasks:
- name: Sorry {{ ansible_user }}
debug:
msg: this won't work
- name: You say '{{ greeting }}'
debug:
var: ansible_user
I experienced the same problem today in one of my Ansible roles and I noticed something interesting.
When I use the set_fact module before I use the vars in the task name, they actually get translated to their correct values.
In this example I wanted to set the password for a remote user:
Notice that I use the vars test_user and user_password that I set as facts before.
- name: Prepare to set user password
set_fact:
user_password: "{{ linux_pass }}"
user_salt: "s0m3s4lt"
test_user: "{{ ansible_user }}"
- name: "Changing password for user {{ test_user }} to {{ user_password }}"
user:
name: "{{ ansible_user }}"
password: "{{ user_password | password_hash('sha512', user_salt) }}"
state: present
shell: /bin/bash
update_password: always
This gives me the following output:
TASK [install : Changing password for user linux to LiNuXuSeRPaSs#]
So this solved my problem.
It might be ugly, but you can somewhat workaround with something like this:
- name: create a virtual host file
debug:
msg: "Some command result"
loop: "{{ [ vhost ] }}"
or
- name: create a virtual host file
debug:
msg: "Some command result"
loop_control:
label: "{{ vhost }}"
loop: [1]
I wouldn't do this in general, but it shows how you can use items or label to give information outside of the command result. While it might not
Source: https://docs.ansible.com/ansible/latest/user_guide/playbooks_loops.html