Variables Conditionals Loops
Overview
Similar with any programming language, variables are used to store information varies with each host.
We can configure variables in the playbook itself using the vars keyword, followed by the key-pairs. Below we have add-2.yaml which adds a thrid server to the /ansible-test/sample-file.txt in server1.
-
name: Add servers to /ansible-test/sample-file.txt on server1
hosts: server1
vars:
anotherserver: 10.0.0.127
tasks:
-
name: Add another server to /ansible-test/sample-file.txt
lineinfile:
path: ~/ansible-test/sample-file.txt
line: 'server {{anotherserver}}'
Checking the /ansible-test/sample-file.txt in server1 again, you should now see a third server:
eden@server1 ansible-test$ cat sample-file.txt
server 10.0.0.125
server 10.0.0.126
server 10.0.0.127
Variables can also be created on a separate file and this separate file can just be referenced by the playbook. This will be further discussed on the section for host_vars and group_vars.
CONDITIONALS
Here we have two playbooks that does the same thing but on different OS - install NGINX. Both install packages differently Debian distros use apt while RedHat uses yum.
Beside the package manager, you'll also have to specify different hosts for each one. Having two playbooks is alright but it can be done altogether using just one. This is where conditionals comes into play.
We can do a single playbook with two tasks and then we specify the conditions under each task.
-
name: Install NGINX on Linux machines
hosts: all
tasks:
-
name: Install NGINX on Debian
apt:
name: nginx
state: present
when: ansible_os_family == "Debian" and
ansible_distribution_version == "18.04"
-
name:
yum:
name: nginx
state: present
when: ansible_os_family == "RedHat" or
ansible_os_family == "SUSE"
Notice that for the conditional statements, we use the keyword when. We also use logical operators and, or, and the '=='. In addition to this, we also specify that apt is to be used for the first task while yum on the second one.
We use the and operator to tell Ansible to make sure both specified conditions are fulfilled. In this case, nginx will only be installed if the OS is Debian and the version on that machine is 18.04.
when: ansible_os_family == "Debian" and
ansible_distribution_version == "18.04"
On the other hand, we use or if atleast one of the conditions is required to be true, and not both.
when: ansible_os_family == "RedHat" or
ansible_os_family == "SUSE"
CONDITIONALS IN LOOPS We can also use conditions on loops. For the example below, we will be installign different services on a Debian machine. We can set the packages as a variable.
-
name: Install services
hosts: all
vars:
packages:
-
name: nginx
required: true
-
name: mysql
required: true
-
name: apache
required: false
task:
-
name: Install "{{ item.name }}" on Debian
apt:
name: {{ item.name }}
state: present
when: item.required == true
loop: "{{ packages }}"
The way I understood this is by reading the YAML from bottom to top. The first point we have here is we are looping or iterating over the packages.
loop: "{{ packages }}"
The packages referred in this instance is a variable which contain a set of packages along with an attribute required - basically the packages is a list. So we are simply iterating over the list and each package on that list is an item
vars:
packages:
-
name: nginx
required: true
-
name: mysql
required: true
-
name: apache
required: false
Now it goes over the first item, which is an nginx. It check if the attribute required is set to true, in which case, it will proceed with the task and install nginx using yum. Otherwise, it won't do anything. It repeats the process onto the second item, the third item, until it reaches the end of the list.
task:
-
name: Install "{{ item.name }}" on Debian
apt:
name: {{ item.name }}
state: present
when: item.required == true
CONDITIONALS AND REGISTER We can also based the conditions on the output of the previous task. We can save the output or returned value of the preceeding task, and use that as basis of the condition. To store the output of the previous task, we use the register followed by the variable we want to use.
In the example below, we have a playbook which checks the status of httpd. If the service went down, it shoudl send an email.
-
name: Check status of httpd service and send email if down.
hosts: server1
tasks:
-
name: Check status of httpd
command: service httpd status
register: status_var
-
name: Send email if down
mail:
to: admin@corp.com
subject: Service Alert
body: Hi, HTTPD service is down.
when: status_var.stdout.find('down') != -1
The first task is first executed. It will run the service httpd status command on ht e linux machine and the output will be stored in the variable status-var which will be used in the second task.
-
name: Check status of httpd
command: service httpd status
register: status_var
Now for the second task, we try to read this from the bottom again. The condition here is to check the variable status_var and find for the word "down". If the word in not found in the variable, it wll return a value of -1.
Otherwise, the value will not be -1 and it will proceed with the task of sending the email.
-
name: Send email if down
mail:
to: admin@corp.com
subject: Service Alert
body: Hi, HTTPD service is down.
when: status_var.stdout.find('down') != -1
LOOPS
In the create-user.yaml below, we'll be using the user module to create users on the target machines. As you can see, this is not efficient since we have multiple tasks that just does the same thing.
-
name: Create users
hosts: server1
tasks:
-
user: name=john state=present
-
user: name=jane state=present
-
user: name=jeff state=present
-
user: name=sam state=present
-
user: name=ron state=present
-
user: name=mack state=present
So we modified the create-user.yaml to include just one task and loop over the list of users.
-
name: Create users
hosts: server1
tasks:
-
user: name="{{ item }}" state=present
loop:
- john
- jane
- jeff
- sam
- ron
- mack
Now let's say we want to create a user and a corresponding user ID. This would mean we have to loop over a set of key-pair values. For this one, we can use dictionaries. We have created a second playbook below, create-user-id.yaml which does exactly that.
-
name: Create user with corresponding user ID
hosts: server1
tasks:
-
user:
name: "{{ item.name }}"
uid: "{{ item.uid }}"
state: present
loop:
-
name: john
uid: 1000
-
name: jane
uid: 1001
-
name: jeff
uid: 1002
-
name: sam
uid: 1003
-
name: ron
uid: 1004
-
name: mack
uid: 1005
IN another example, create-user-id-with.yaml, we see another way to write loops using with_items.
-
name:
hosts:
tasks:
-
user:
name: "{{ item.name }}"
uid: "{{ item.uid }}"
state: present
with_items:
- john
- jane
- jeff
- sam
- ron
- mack
The with_ can also be useful in other instances, like handling files, working on a set of URLs, or connecting to multiple database.
Note that each one used their own keyword. There are a lot more of these, but below are just some example.
- with_files
- with_url
- with_mongodb
