The change happened in Train, time to move on.

Change-Id: Ie58265284b2e6b4b30b24fc2f22dd4f5eec05d5b

The apparmor_parser actually doesn't remove the file or doesn't create
the symlink in '/etc/apparmor.d/disable' itself so the next run of the
baremetal role will fail with the error "Unable to remove "libvirtd".
Even more after reboot, the profile is still active. We need to
disable the profile completly ourselves. This change fixes the
idempotents of the baremetal role.

Closes-Bug: #1960302
Change-Id: I162e417387393e806886b1c9ea8053b89778b4d1
Signed-off-by: Maksim Malchuk <maksim.malchuk@gmail.com>

The default configuration was changed to use the advanced cache pool in
keystonemiddleware 9.3.0 (Xena release) [1].

This reverts commit 5a52d8e4 (except the
release note).

[1] https://review.opendev.org/c/openstack/keystonemiddleware/+/773939

Change-Id: I290d0a81c57c189b6eb62fc3eee3ed19f441671b

parted hangs waiting for user input (see examples below)
on Debian and Ubuntu nodes which have created a cinder
volume on lvm, causing POST_FAILURE of the entire CI job.
Zun (Cinder iSCSI LVM) jobs are affected.
parted seemingly tries to interpret contents of the created
volume and fails miserably.
Since there is no reason why we would need to see the output
of parted specifically, this patch is switching to use
lsblk to simply list visible block devices.
Along with the rest of the commands, this should be just
the right level of detail.
And we avoid having parted interpret internals of otherwise
opaque block devices.

Example issues:

Warning: Not all of the space available to
/dev/mapper/cinder--volumes-cinder--volumes--pool appears to be used, you can
fix the GPT to use all of the space (an extra 9732096 blocks) or continue with
the current setting?
Fix/Ignore?

Warning: Not all of the space available to
/dev/mapper/cinder--volumes-cinder--volumes--pool-tpool appears to be used, you
can fix the GPT to use all of the space (an extra 9732096 blocks) or continue
with the current setting?
Fix/Ignore?

Warning: Not all of the space available to
/dev/mapper/cinder--volumes-cinder--volumes--pool_tdata appears to be used, you
can fix the GPT to use all of the space (an extra 9732096 blocks) or continue
with the current setting?
Fix/Ignore?

Change-Id: I7beecf2dd6c49c8934722cf22efa74e920ecb060

Enable libvirt TLS in CI jobs with TLS enabled.

Uses the new functionality of the certificates command to generate
certificates for both libvirt client and server (added in
I1bde9fa018f66037aec82dc74c61ad1f477a7c12).

Change-Id: Ica304685b043f699799ccee6c9c2fbcf968888db

Adds support to the 'kolla-ansible certificates' command for generating
certificates for libvirt TLS, when libvirt_tls is true. The same
certificate and key are used for the libvirt client and server.

The certificates use the same root CA as the other generated
certificates, and are written to
{{ node_custom_config }}/nova/nova-libvirt/, ready to be picked up by
nova-libvirt and nova-compute.

Change-Id: I1bde9fa018f66037aec82dc74c61ad1f477a7c12

Enables zun to access cinder volumes when cinder is configured to use
external ceph.
Copies ceph config file and ceph cinder keyring to /etc/ceph in
zun_compute container.

Closes-Bug: 1848934
Change-Id: Ie56868d5e9ed37a9274b8cbe65895f3634b895c8

This gets rid of one of the steps in the authentication flow.

Closes-Bug: 1930055
Change-Id: I4ed4651b55a912f1d9aec7277bae6bb4776f1e0a

This is required to use, e.g., Cinder iSCSI backend with Glance.

Closes-Bug: #1959663
Change-Id: I6d5e0e4ab70922a772f3c82e914f9e7d37cf0318

Custom themes support for horizon deployment.
 - horizon role task added for copying theme files
 - added theme copying into the container in templates
 - docs and reno updated

Change-Id: If9982c8e18be31772cb031ef72b7eebd4d768be5
Co-Authored-By: Jakub Darmach <jakub@stackhpc.com>
Depends-On: https://review.opendev.org/c/openstack/kayobe/+/824565
Depends-On: https://review.opendev.org/c/openstack/kolla/+/826672

This fixes a bug in registering identity providers

The bug was caused by a missing `=` in the openstack command

Add the missing `=` after `--os-user-domain-name`

Closes-Bug: #1959022
Change-Id: I73f80cd2c81a3944de0933e60f5768956a1a3b70

Overrides default fluentd buffer config to stop log files from using
datestamped filenames, allowing logrotate to manage them.

Closes-Bug: #1940118
Change-Id: I40c4e209470d21e0a02fd447fb628acfdae9fa9d

The value of node_custom_config should is {{ node_config }}/config,
when specified using --configdir

Change-Id: I076b7d2c8980ddd3baa28f998f84a6b7005dc352

There are multiple Neutron extensions that can be used for DNS
integration. The "dns" extension has only minimum functionality, for
most deployments either "dns_domain_ports" or
"subnet_dns_publish_fixed_ip" would be used, with the latter being a
superset of the two others [0].

[0] https://docs.openstack.org/neutron/xena/admin/config-dns-int-ext-serv.html#use-case-3-ports-are-published-directly-in-the-external-dns-service



Signed-off-by: Dr. Jens Harbott <harbott@osism.tech>
Change-Id: I1056bddfd49265ffe21b4d000898d72bba3ebfad