+++ This bug was initially created as a clone of Bug #1556946 +++
WaitCondition uses obsolete method to generate the passwords (uuid.uuid4().hex) [1]. As a result, generated passwords don't contain special characters or upper case letters, so it is impossible to validate them against keyston's strong password policies.
Steps to reproduce:
Enforce strong password policies in keystone. Use Heat templates with WaitConditionHandle.
Expected result:
Heat templates successfully deployed.
Actual result:
Heat fails with the following output:
resource_type: OS::Heat::WaitConditionHandle
physical_resource_id:
status: CREATE_FAILED
status_reason: |
BadRequest: resources.saltmaster_wait_handle: The password does not match the requirements: Password must contain 8 characters and at least one uppercase letter, one lowercase letter, one number and one unique symbol.. (HTTP 400) (Request-ID: req-4e0d89b1-1ad2-420c-86a2-3bdbdf0a3581)
PS. I have checked the heat code and it looks like we already implemented password generator [2], [3]. So we basically should just change the password generation methods.
[1] https://github.com/openstack/heat/blob/master/heat/engine/resources/openstack/heat/wait_condition_handle.py#L138
[2] https://blueprints.launchpad.net/heat/+spec/random-string-resource
[3] https://github.com/openstack/heat/blob/master/heat/engine/resources/openstack/heat/random_string.py
--- Additional comment from Zane Bitter on 2018-03-19 13:17:04 EDT ---
There's a related-but-not-identical bug opened upstream too: https://bugs.launchpad.net/heat/+bug/1666129
There was a patch proposed for that one that involved retrying in a loop until you found a successful password, but the author unfortunately abandoned it after I suggested using the password generator from RandomString just as you did.
I agree that that's the right approach, especially now that https://bugs.launchpad.net/heat/+bug/1745931 is fixed. It will involve some refactoring, however, so it remains to be seen how much of a problem that poses for backporting.
--- Additional comment from Zane Bitter on 2018-03-22 10:01:58 EDT ---
I proposed patches upstream. It should be possible to eventually get these backported (the refactoring one includes a security improvement for OS::Heat::RandomString, which makes it a candidate for stable branch backporting even upstream) once they've been reviewed. In the meantime, the workaround posted in the upstream bug is a good one: modifying the Keystone password_regex to allow passwords without special characters if they are long enough:
"< your original regex >|^(?=.*?[a-zA-Z])(?=.*?[0-9]).{30,}$"