4.2. RHEA-2016:1599 — Red Hat OpenStack Platform 9 director Release Candidate Advisory

The bugs contained in this section are addressed by advisory RHEA-2016:1599. Further information about this advisory is available at https://access.redhat.com/errata/RHEA-2016:1599.html.

4.2.1. instack-undercloud

The ironic-inspector was not included in the Keystone endpoint catalog in the Undercloud. This meant you could not discover the API endpoints for ironic-inspector through the catalog. This fix adds the ironic-inspector API endpoints to the Keystone catalog. You can now discover the ironic-inspector API endpoints through the catalog.
Slow environments experienced timeouts when Glance tried to communicate with Swift as a backend. This caused some Glance operations, such as image uploads, to fail. This fix increases the Swift proxy server's default node_timeout value to 60 seconds. This increases the reliability of Glance image uploads on slow environments using Swift as an image storage backend.

4.2.2. ipxe

The ipxe-bootimages RPM was not included in the repository for the director. This caused failure during director installation. This update adds the package to the director's repository. The RPM is now included as a part of the director's installation.

4.2.3. openstack-tripleo-heat-templates

Almost all the Overcloud Pacemaker resources depended on the Keystone resource. This meant restarting the Keystone resource after a configuration change would restart all dependent resources, which was disruptive. This fix introduces a fake openstack-core that the Overcloud Pacemaker resources (including keystone) use as a dependency. This means restarting the Keystone resource no longer causes any disruption to other services.
The ManagementNetValueSpecs parameter used the wrong type (string) in the director's Heat templates. This caused Overcloud deployments containing the Management network to fail with the following error:

"Property error: resources.ManagementNetwork.properties.ManagementNetValueSpecs: Value must be a string".

This fix changes the data type of ManagementNetValueSpecs from string to json. The error no longer appears.
In previous releases of Heat, domain resources were created before /etc/heat/heat.conf was configured on the overcloud. However, domain resources depended on settings from that file; as such, these resources were not created correctly, preventing users from creating heat stacks. Users had to manually restart the Pacemaker Heat Engine resource to work around the problem.

This release corrects the sequence of steps for deploying the Heat service, thereby fixing the problem.
Old versions of the puppet-ceph-external.yaml Heat environment file are not supported anymore. A newer version of this file is shipped with the 9.0 templates and must be used instead. Any customizations made to the old puppet-ceph-external.yaml file must be ported into the newer version.
This patch updates the Help URL on the dashboard. The URL now points to the Official Red Hat OpenStack Platform documentation instead of the OpenStack upstream documentation.
The default newtork-isolation.yaml file contained values that conflicted with default network-management.yaml file in the director's Heat templates. If you included the network-isolation.yaml file after the network-management.yaml when creating an Overcloud, the deployment deactivates the Management network. This update refactors these files to avoid conflicts. Now the deployment processes both the network-management.yaml and network-isolationl.yaml in any order without conflict.
The Overcloud uses a fake openstack-core resource to replace keystone (now accessed through httpd). However, the Overcloud did not create openstack-core when using an external load balancer. This caused puppet to fail when creating Pacemaker constraints that required openstack-core. This fix ensures creation of the openstack-core resource for any deployment configuration. The deployment succeeds and creates the needed constraints.
OpenStack Platform 9 deployments require an additional CephX key for the "client.openstack" user. However, the director's command line client does not generate this key for existing deployments and updates the "ceph.openstack" keyring with an empty secret. Before upgrading, generate a new CephX key and pass it to the deployment using the CephClientKey parameter in an environment file. For example:

    CephClientKey: 'my_cephx_key'

Generate the new key with the following command:

$ ceph-authtool --gen-print-key

4.2.4. os-collect-config

The "os-collect-config" service on the Overcloud restarted on an RPM update. This caused Overcloud updates to fail. This fix changes the behavior so that "os-collect-config" does not restart on an RPM update. The Overcloud updates now succeed after an update of "os-collect-config". Note that "os-collect-config" gracefully restarts itself when "os-refresh-config" runs, so the restart on update is not required.