Red Hat Bugzilla – Bug 1459729
Compute resource for Openstack uses adminURL instead of publicURL for accessing Openstack
Last modified: 2017-11-22 11:01:11 EST
Description of problem:
When adding an openstack compute resource to Satellite, it will throw errors connecting to some Openstack resources (such as cinder) because it defaults to connecting to Openstack services on the adminURL instead of the publicURL. This is because the ruby fog library it uses defaults to adminURL. See
This causes issues because by default, RHOS and RDO TripleO installations always put adminURLs on internal networks that aren't publicly accessible. This is valid because for most non-keystone services, the public and admin urls have the same functionality, and the adminURL is not needed.
Version-Release number of selected component (if applicable):
Reproducable easily with Satellite 6.2 against RHOS 11
Steps to Reproduce:
1. Do a standard RHOS 11 installation with TripleO
2. Configure compute provider in satellite to point to RHOS 11 installation
3. Attempt to create an instance backed onto a cinder volume
Get an error failing to connect to (cinder internal url)
instance is created successfully
We need to update
to expose the option openstack_endpoint_type to the user (or default it to publicURL)
This was fixed through projects.theforeman.org/issues/12054 - now we default to publicURL.
Closing as POST and duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=1513932 since the fix is the same.
*** This bug has been marked as a duplicate of bug 1513932 ***