Accelerate Ansible

Question 1

What is SSH multiplexing

Answer 1

Reuse of TCP connections to execute many SSH sessions. Traditionally, each SSH session establishes a separate TCP connections, which incures negotiation penalty.

By using SSH multiplexing, playbooks are able to reuse the TCP connection

Question 2

What is ControlPersist?

Answer 2

Same as SSH multiplexing

Question 3

How do you check if ssh multiplexing is enabled for a site myserver.example.com?

Answer 3

ssh -o check ubuntu@myserver.example.com

Question 4

How do you termine the master connection for a site myserver.example.com

Answer 4

ssh -o exit ubuntu@myserver.example.com

Question 5

What are the three parameters that can be configured for customizing multiplexing behavior?

Answer 5

• ControlMaster: tell if ssh multiplexing should be enabled
• ControlPath: tells where tos tore the socket file for esach node
• ControlPersist: how long the master connection should be kept idle

Question 6

When talking about ssh multiplexing, explain what is a master connection.

Answer 6

Master connection is the reusable connection that TCP creates to talk to a host when SSH multiplexing is enabled.

Question 7

Despite enabling SSH multiplexing, TCP might sometimes not setup a master connection. Explain why?

Answer 7

SSH multiplexing opens a file to represent each master connection. It uses the hostname as the name of the file. If this filename is too long, the operating system might reject its creation; causing TCP to error in the creation of the master connection and therefore falling back to creating a new socket each time.

Question 8

In what situations can it make sense to disable fact gathering and why?

Answer 8

if my playbook doesn’t use facts, I can disable fact gathering to speed up the playbook execution.

Question 9

How do you disable fact gathering for one playbook?

Answer 9

gather_facts: false

Question 10

How do I disable fact gathering for all my playbooks?

Answer 10

In ansible.cfg default section:

[defaults]
gathering=explicit

Question 11

How can I speed up my playbook fact gathering process?

Answer 11

Use fact caching

Question 12

How do you enable fact caching?

Answer 12

in ansible.cfg

[defaults]
gathering=smart
fact_caching=jsonfile

Question 13

How do you specify fact caching timeout?

Answer 13

In ansible.cg (in seconds)
[defaults]
fact_caching_timeout= 300

Question 14

What are the fact caching implementations supported by ansible?

Answer 14

redis
memcached
json

Question 15

How do I specify which fact caching implementation Ansible should use?

Answer 15

In ansible.cfg,

[defaults]
fact_caching=jsonfile (or redis or memcached)

Question 16

What is the default ansible fact caching implementation?

Answer 16

There is no default fact caching implementation. if you do not specify one, Ansible will not cache facts.

Question 17

I want to use json for caching facts. How do I specify where these files should be stored?

Answer 17

in ansible.cfg

[defaults]
fact_aching_connection=/tmp/ansible_fact?cache

Question 18

I want to use memcached for fact caching, how do I specify the address and port of the memcached instance to be used by ansible?

Answer 18

memcahed instance should be installed locally to the control node.

Question 19

I want to use redis for fact caching, how do I specify the address and port of the memcached instance to be used by ansible?

Answer 19

redis instance should be installed locally to the control node.

Question 20

If I have an inventory of 500 nodes, will Ansible connect to all of them in parallel to execute my playbook?

Answer 20

The short answer is no. While Ansible connects to node in parallel, the number of nodes ansible connects to simultaneously is limited. This limit can be defined in ansible.cfg

[defaults]
forks=20

Question 21

Explain:

[defaults]
forks=20

Answer 21

This is a configuration that goes into ansible.cfg to define how many managed nodes ansible.cfg should connect to in paralle.

Question 22

What is the use of async in Ansible?

Answer 22

Asynchronous tasks are useful in following scenarios:

• task that would take so long that the tcp connection would timeout before the task is completed
• ## start a second task before the the previous one is completed

Question 23

What happens when the socket timeout is shorter than the time it takes to execute a task?

Answer 23

The socket may timeout causing ansible to report the task as failed. This can be avoided by making the task async.

Question 24

How does Ansible know that an async task is completed?

Answer 24

it periodically polls the node to check whether the tsk is complete, sleeping between checks

Question 25

If I use async, how do I capture the result of the task execution?

Answer 25

Use register clause to capture the async result

Question 26

Explain pipelining

Answer 26

Instead of copying files, running them on the remote server, then removing them, pipelining sends and executes commands directly over the SSH connection