module_utils/ec2.py Do we need the AWS_REGION list

Archives Ansible Developer
,
1

Fedora Infrastructure is using openstack with the ec2 compat layer and
thus the region was being set to something local rather than the
official aws regions. It looks like boto allows this via the
BOTO_ENDPOINTS environment variable (or a config file setting).
However, the ansible ec2 modules are additionally doing a check
against a hardcoded list of regions.

It seems like we might want to only depend on the regions that boto is
able to find because boto has less chance to become stale than we do.
So I'm inclined to accept this PR from them to remove the separate
check in ansible for this:

https://github.com/ansible/ansible/pull/10734

I just wanted to check with people who use ec2 more than I do about
ec2 about whether there would be any unwanted side effects from this
change.

Thanks for any feedback (here or preferably in the PR),
-Toshio

2

To do something along these lines has been on my backlog of things to tackle. I think this is a good change. IIRC none of the AWS modules have an dependency on that hardcoded list (to include the dynamic inventory script which looks to the ini file anyways). In the highly unlikely event that this causes a particular module to stop working as intended it would be 1) Relatively easy to fix 2) Something that would have needed to be fixed anyways. 3) Better for the whole of our AWS modules in the long run.