Fixed the issue that caused the Unable to retrieve image list error message
to appear on the Admin > Compute > Instance tab in OpenStack
Horizon.
Fixed the issue with inability to modify access for an existing flavor in
OpenStack Horizon.
Fixed the issue with OpenStack Horizon being unavailable if apache2 was
stopped on a VIP prx node.
Fixed the issue that caused failures of the OpenStack Pike or Queens
deployment in case of Keystone connected to LDAP.
Fixed the following RabbitMQ and Oslo issues:
RabbitMQ crashes
RabbitMQ failing to recover the cluster during the network segmentation
Sporadic message loss causing unreliable functioning of Neutron and some
other services
Fixed the issue with the Deploy - upgrade control VMs Jenkins
pipeline job failing with the Service Unavailable (HTTP 503) error message.
Fixed the issue with some services failing to start in case of the
cinder-backup service enabled in OpenStack Queens.
Fixed the issue with the cinder-backup service failing to start in
OpenStack Queens.
Fixed the issue with libvirt occasionally creating incorrect AppArmor rules.
Fixed the issue with the Deploy - OpenStack Jenkins pipeline job
failing in case of Redis 5.0 package installed.
Fixed the issue with the Apache server-status module that exposed the
server status metrics to an external proxy allowing an unauthenticated user
to access the Server-Status web page in Horizon and gather
sensitive information.
Fixed the issue with the Tempest tests failures when Nova integration with
Barbican is enabled. The Tempest tests for Nova that perform booting from an
instance snapshot and unshelving of a shelved instance are now being skipped
to prevent the Tempest tests failures. For details, see:
MCP Deployment Guide: Deploy Barbican.
Fixed the issue with MCP OpenStack deployments keeping only four log
rotations, which could prevent from investigating the issues in detail. Now,
MCP OpenStack deployments keep logs for 10 days by default.
Fixed the issue with the Keystone catalog containing deprecated Cinder API v1
endpoints, which raised the CinderApiDown alert.
Fixed the issue with the Designate Tempest test failing in case of a small
number of Designate quota zones configured.
Added the capability to set certain HAProxy check parameters to fix the issue
with HTTP checks failing due to the lack of configuration when using SSL for
the Designate, Glance, Heat, Keystone, Nova, Manila, and Neutron services.
Fixed the issue with the Redis server not requiring authentication.
Fixed the issue with Gnocchi containing a significant number of errors in
the metricd logs in the Queens-based OpenStack deployments with
Telemetry.
Fixed the issue with the OVS network configuration occasionally failing to
apply after rebooting a node on the OpenStack Pike or Queens environments
with a custom interfaces configuration. This could cause the Exhausted all
hosts available error for build instances.
Fixed the issue with the ohm port moving to the DOWN state in case of
the gtw01 node reboot, which affected the creation of new load
balancers on the OpenStack environments with Octavia enabled.
Fixed the issue with the Deploy - OpenStack deployment pipeline
job failing when the LDAP integration for Keystone is enabled. The issue
affected the OpenStack Pike and Queens releases.