Customizing Hosts, Runs, Handlers

Question 1

What are possible host patterns?

Answer 1

all, union, intersection, exclusion, wildcard, range, regular expression

Question 2

Explain intersection in host definition and give an example

Answer 2

Apply a task on the intersection of two groups.

hosts: database:&staging

Question 3

Explain union in host definition and give an example

Answer 3

Apply a task on the union of two groups.

hosts: database:staging

Question 4

Explain exclusion in host definition and give an example

Answer 4

Apply a task on the exclusion of a group from another groups.

hosts: database:!staging

Question 5

Explain wildcard in host definition and give an example

Answer 5

Apply a task on set of hosts matching an expression defined with wildcard.

hosts: *.example.com

Question 6

Explain range in host definition and give an example

Answer 6

Apply a task on set of hosts defined by range

hosts: web[1:20].example.com

Question 7

Explain regex in host definition and give an example

Answer 7

Apply a task on set of hosts defined by regular expression

hosts: ~web\d+.example.(org|com)

Question 8

Sketch a command that would run a playbook.yml against a group of hosts web as opposed to the whole invetory.

Answer 8

ansible-playbook -l web playbook.yml

Question 9

Sketch a command that would run a playbook.yml against a group of hosts using filters.

Answer 9

ansible-playbook -l web:staging playbook.yml

Question 10

What clause allows you to run a command on the control machine?

Answer 10

local_action

Question 11

Give an example scenario where running a command on the control host makes sense and provide the correspdoning ansible task.

Answer 11

We restarted the managed host and we have to wait for it to be back before we do other configuraions. In this case, the wait will be on the control node.

-name: wait for managed host to be back
local_action: wait_for port=22 host=”{{ inventory_hostname}}” search_regex=OpenSSH

Question 12

What are some use cases where you might want to run a command on a node other than the control and the managed node?

Answer 12

1) You have to add the managed node to a load balancer

2) You have to configure an alerting system to watch the managed node

Question 13

What is the key clause that enables running a command on a node other than the control and the managed nodes?

Answer 13

delegate_to

Question 14

Provide an example task of running a command on a node other than the control and the managed nodes?

Answer 14

• name: enable alerts
  nagios: action=enable_alerts service=web host={{ inventory_hostname}}
  delegate_to: nagios.example.com

Question 15

What is the default execution mode: one node after anorher or all nodes in parallel?

Answer 15

all nodes in parallel.

Question 16

Cite an example where running ansible on managed node in sequence would make sense as opposed to running in parallel.

Answer 16

Upgrading a production application where many nodes are behing a LB. Becase don’t need the service interruption, we would upgrade one node at a time.

Question 17

What clause do you use to run the playbook on one node at a time?

Answer 17

serial:1

Question 18

What clause do you use to run the playbook on 20% of node at a time?

Answer 18

serial: 20%

Question 19

I have 5000 nodes in my inventory. I want to try the playbook on one node first and then run it on 10% of nodes at a time. what is the right construct for this?

Answer 19

serial:

- 1
- 20%

Question 20

Explain max_fail_percentage

Answer 20

Allows you to specify the number of nodes that are allowed to fail when running the playbook. The value is in percentage. Beyond the specified percentage, Ansible will stop running the playbook.

Question 21

What is the clause for running a playbook only once?

Answer 21

run_once: true

Question 22

What scenario can it be useful to run the playbook only once?

Answer 22

database initialization or sometimes local actions

Question 23

What are example of ansible running strategies?

Answer 23

linear, free

Question 24

What is the default Ansible running strategy?

Answer 24

linear

Question 25

What is a running strategy?

Answer 25

Strategy is a mechanism that determines in which order Ansible will run the tasks across the nodes. For instance execute the same task on all nodes, wait for it to complete before proceeding to the next task, or do not wait for all tasks to complete and proceed to the next task once a node is node running the previous task.

Question 26

Explain ansible linear strategy.

Answer 26

Running strategy wherein ansible executes the same task on all nodes, wait for the task to complete on each of the nodes before proceeding to the next task.

Question 27

Explain ansible free strategy.

Answer 27

Running strategy wherein when executing a tasks on a node Ansible does not wait for other nodes to have completed this task. Any node that is done executing a task can proceed to the next task in the play.

Question 28

In what cases can it be useful to employ the free running strategy?

Answer 28

When nodes have different hardware specifications such that some may be faster than others. Network latency can also be a parameter.

Question 29

It is possible to execute a handler many times in one play. Explain how.

Answer 29

The same handler can be triggered from pre_tasks, tasks, or post_tasks sections. Each of these sections will trigger a separate execution of the handler.

Question 30

What is the scope of a handler?

Answer 30

The scope of a handler is the play. You cannot nofify a handler beyond a play.

Question 31

What problem do you see in listing the names of handlers in a notify clause?

Answer 31

hard coupling of handlers and tasks. This can get cumbersome if the notify includes many handlers.

Question 32

How can you solve the issue of hard coupling between tasks and handlers?

Answer 32

Make handlers to listen to changes as opposed to being notified.

handlers:

• name: restart postfix
  service: name=postfix state=restarted
  listen: postfix config changed

Question 33

What is one of the fundamental benefits of handlers listen?

Answer 33

handlers across roles and solving role dependencies. In cases where a role has to notify the depending roles of any changes, this introduces unwanted strong coupling. E

Question 34

What is the default behavior if a notify is defined without handler?

Answer 34

Ansible will complain about notifying missing handler.

Question 35

Explain the error_on_missing_handler flag.

Answer 35

When set to true, Ansible will complain about notifying missing handler. When set to false, Ansible will not complain.