Role setting remote locale to same as local locale instead of var defined one about ansible-role-locale

[robertdebock]

Don't worry, I learn from these kinds of issues too. So far; your issue are certainly interesting.

Thanks for following up!

[robertdebock]

It's not easy to verify (molecule/default/verify.yml) this issue, bacause Ansible sets a default language.

I have tried to verify better however.

For your playbook, where does the de_AT.UTF-8 value come from? It does not come from the playbook (that uses the role) and it's also no the default... Is it maybe set for a specific user? Could it be that SSH is passing that variable from your machine to the managed server?

[Migsi]

It's not easy to verify (molecule/default/verify.yml) this issue, bacause Ansible sets a default language.

This is probably set to de_AT.UTF-8, as that's the locale my notebook (the controller) is set to.

For your playbook, where does the de_AT.UTF-8 value come from? It does not come from the playbook (that uses the role) and it's also no the default... Is it maybe set for a specific user? Could it be that SSH is passing that variable from your machine to the managed server?

As I explained above, my notebook is set to this locale. As you suspected, I checked and noticed my ssh config is indeed forwarding the local locale. The server is configured to accept it. Are you suggesting to either disable reception on the server side or forwarding on my local machine and try again?
I'd prefer to leave these configs as it, as I'd have to check quite a few other tasks if they might be impacted by such changes. Just to get this right, what is the actual impact of ssh forwarding the locale setting while using this role? Shouldn't the locale change by the role be unaffected by any "outside" properties? That was my expectation when I started using the role and it worked fine, but please correct me if I got this wrong.

The overall behavior is weird though, as none of the related configs changed over time. The only thing that changed is that I updated ansible (to core 2.11.3) and the role to its most recent version (3.1.0). That's when the play broke and I started to see the reported behavior. That is why I suspected changes within ansible to be responsible, allthough I was unable to verify that (yet).

[Migsi]

I guess I owe you an apology for bothering you again with issues that aren't actual issues. I just rechecked all locale related settings to make absolutely sure I wasn't misinterpreting something, but appearently I was. 🤦‍♂️

Your hint about ssh accepting the LC_* from the host logging pointed me to the curlprit. The servers default locale was always properly set to the value defined by locale_lang but as sshd is accepting remote locales it obviously looked to me as if the server was using the same locale as I do. When logging in via VNC everything was as it should always have been and thus the role never had an issue.

I'm still wondering though, why I didn't see this behavior before the above mentioned updates of ansible and the role. 🤔

I'll close the issue with a big thank you for taking your time helping me figuring out my (stupid) mistakes! 😅 I'll also open up a pull request adding information about this behavior to your readme, hoping it will prevent others from facing the same "problem".