Help understanding how the limit flag works with the ansible-playbook command

[Daniel Richardson]

I ran into what seemed to me like a strange problem today, but I might just lack the understanding to know what was happening. 

Let's say you have a YAML file called ec2.yml to run against that looks like this:

---

- hosts: tag_ansible_host1
  roles:
     - role1
     - role2

- hosts: tag_ansible_host2
  roles:
     - role2

Both of these hosts are at ec2 and you are using the ec2 dynamic inventory script. Both hosts are also tagged with ansible:host1 and ansible:host2.

I was in a similar situation today and ran a similar command to the following:

ansible-playbook -l tag_ansible_host2 ec2.yml

What I expected to happen was that ansible would only run role2 on both hosts since I specified the limit flag. However, what happened was that ansible began running role1 and role2 on both hosts. I suspect either I am just missing something about how the limit flag works. Can anyone help me understand what I did wrong?

[Brian Coca]

limit filters the available host list, since both hosts have the host2
tag, it does not modify the list of hosts in the play.

it reads like this once resolved

- hosts: "tag_ansible_hosts1:&tag_ansible_host2"
...

- hosts: "tag_ansible_hosts2:&tag_ansible_hosts2"

It does not limit to an existing play declaration, but imposes
additional limits to those. What you seem to want is play tags.

- hosts: tag_ansible_host1
tags:
- host1

roles:
- role1
- role2

- hosts: tag_ansible_host2

tags:
- host2
roles:
- role2


and then run ansible with -t 'host2'



--
Brian Coca

[Daniel Richardson]

Thanks so much Brian. That makes a lot of sense now.