Precendence: why playbook level variables win over inventory ones?

[Stanislav German-Evtushenko]

Documentation (see 4-7, 9, 10 from the list of priorities)  says that playbook level variables override an inventory ones. Can somebody explain why this is the case? To me this is counter-intuitive and keeps leading to mistakes.

Intuitive way would be:

1) role defaults (specified within the role itself)
                  ⇩
2) infrastructure defaults, e.g. datacenter level (specified in group_vars or host_vars on the playbook level)
                  ⇩
3) inventory defaults, e.g. availability zone, or a specific project, etc (specified in group_vars or host_vars on inventory level)

However the fact is that 2 always overrides 3. How was it supposed to be used?

[James Cassell]

On Fri, Oct 18, 2019, at 11:53 PM, Stanislav German-Evtushenko wrote:
> Documentation
> <https://docs.ansible.com/ansible/2.8/user_guide/playbooks_variables.html#variable-precedence-where-should-i-put-a-variable> (see 4-7, 9, 10 from the list of priorities) says that playbook level variables override an inventory ones. Can somebody explain why this is the case? To me this is counter-intuitive and keeps leading to mistakes.

I agree that it's not intuitive... It's also backwards from what I would expect.

You can override the precedence order locally: https://docs.ansible.com/ansible/latest/reference_appendices/config.html#variable-precedence

I choose not to do that, and instead forgo separating inventory from playbook group_vars.

I, too, would be interested in an explanation of why the order is what it is. I've been bitten by it in the past.


V/r,
James Cassell

[Stanislav German-Evtushenko]

I agree that it's not intuitive... It's also backwards from what I would expect.

You can override the precedence order locally: https://docs.ansible.com/ansible/latest/reference_appendices/config.html#variable-precedence

I choose not to do that, and instead forgo separating inventory from playbook group_vars.

I, too, would be interested in an explanation of why the order is what it is.  I've been bitten by it in the past.

 

Ansible is supposed to be simple and overriding precedence would make it a way more complicated. Changing the default behaviour (with an option for backwards compatibility) would make it more logical. Any devs here?

[Matt Martz]

A few things to note here.

4-10 are all talking about host related vars.  It's just that group_vars and host_vars can live at 2 locations:

1. relative to the inventory file 

2. relative to the playbook file

It is assumed that if you put host_vars and group_vars relative to the playbook, that they are more specific to that playbook, as such they override host_vars and groups_vars that are relative to the inventory file.

--
You received this message because you are subscribed to the Google Groups "Ansible Project" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ansible-proje...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ansible-project/189f4180-9837-43a2-b8cf-57eec9ed61bf%40googlegroups.com.

--

Matt Martz
@sivel
sivel.net

[Stanislav German-Evtushenko]

> It is assumed that if you put host_vars and group_vars relative to the playbook, that they are more specific to that playbook, as such they override host_vars and groups_vars that are relative to the inventory file.

I'm having hard time imagining how this can be useful and when somebody would want to override inventory variables by the play ones. What would be the use case? What would be the files structure?