"Smoke tests" for barbican, cinder, glance and keystone have been removed as discussed in PTG April 2022.
Signed-off-by: Tim Beermann <beermann@osism.tech>
Change-Id: I613287a31e0ea6aede070e7e9c519ab2f5f182bd
Add an enable_cinder_backend_pure_iscsi and
enable_cinder_backend_pure_fc options to etc/kolla/globals.yml
to enable use of the FlashArray backend.
Update the documentation to include a section on configuring
Cinder with the FlashArray.
Implements: blueprint pure-cinder-driver
Change-Id: I464733f1322237321ed1ffff8636cf30bd1cbb38
This is in the hope to avoid random "Unable to validate token" in
AIO upgrade jobs which restart the Keystone container while
other services are trying to use it (most notably Placement but
also Nova and Neutron).
Change-Id: Ie6a95c3ca84df82ca8463cf76054e125cb1ffff1
Facts define the group key to judge in incloud roles,
remove when statement does not execute to speed up execution
Partially-Implements: blueprint performance-improvements
Change-Id: If22255f1adc07ab16b46f8ad1280efdf7d713d28
With the parameter bifrost_deploy_verbosity it is possible
to set the verbosity of the bootstrap of Bifrost.
This makes it possible to reduce verbosity when running
/bifrost/playbooks/install.yaml if needed.
Change-Id: I5815220f2193a492ae7e1f63443075790ae7aaef
Zed dropped binary images so we need to check does migration from binary
images works.
Dropped all other CI jobs
Change-Id: Id2d7ca01baca8e17a9780345e705ffb6cbffd8a7
We run some nova tasks once per cell, using a condition to match a
single host in the cell. In other similar tasks, we use run_once, which
will fail all hosts if the task fails. Typically these tasks are
critical, and that is desirable. However, with the approach used in
nova-cell to support multiple cells, if a once-per-cell task fails, then
other hosts will continue to execute, which could lead to unexpected
results.
This change adds any_errors_fatal to the plays or blocks that run these
tasks.
Closes-Bug: #1948694
Change-Id: I2a5871ccd4e8198171ef3239ce95f475f3e4b051
When running in virtual environment a bootstrapping task will
install 'docker' python module in the environment. After that
prechecks and deploy phase will not find the module because
their 'ansible interpreter' is host wide. With this fix it is
advised to set an interpreter to the one in virtual environment
Closes-Bug: #1969812
Change-Id: I1bb6a3af3fa6346fff0a262e0c53d12c544f46db
For Swift, CI was still using DockerHub, directly.
The issue with DockerHub itself is that we publish there only
weekly.
The issue with not using infra's mirror is that we waste precious
resources.
Time to end that drama.
Change-Id: I928090c24f0eb24f8fab9b7412b58e854246fe80
There are no 5.* versions of ansible for CentOS 8 Stream yet.
One should use pip install 'ansible>=4,<6' according to the
current documentation.
Closes-Bug: #1969617
Change-Id: Ie4f502f955dac5ae9ee8ddb4779c2fa2e26840d9
This uses the same approach as the mariadb role (and others).
Closes-Bug: #1928193
Co-Authored-By: John Garbutt <johng@stackhpc.com>
Change-Id: I79a7a8c80327cfd9ef31d17fe71f450a181a638c
