We need an empty file because as bootstrap-host runs, it expects
that a $service.yml.aio exists for each service running.
We run Rally in the utility container therefore this patch just
adds an empty file to stop Ansible from crashing.
Change-Id: Ia645851a23595a5a988bd0073f90fe6ef361d521
The aio_metal.yml.example had some missing and extra containers
which meant that their metal jobs still built containers.
This patch fixes that.
Change-Id: If76add769defea6b68843f2d09bd6eee4dc84741
CentOS has slowed down significantly. It takes around 2h30m and
a lot of successful runs are failing with a few minutes short.
Let's give it a lot more space to land code, but let's also take
this time to figure out what's going on.
Change-Id: I53c5757e45d103365c446a02da64c9cc659a4907
We added changes to gating config to use different config if we
have nspawn however it is non-functional and causing issues for
the metal builds.
Let's restore the value to br-storage always until we resolve
these issues.
Change-Id: Ibffc81eeaa93d852e86ebe35afeae3c5a38dc43a
This change deploys Swift if Ironic is requested to be deployed
as it has certain dependencies on it.
Change-Id: I6989a7ebc4d35216ec1a3f802204511fcdf360df
These scenarios will be defined in their respective role
repositories instead. This reduces the changes to the job
configuration in this repo, which reduces the merge
conflicts when trying to add new scenarios.
Change-Id: Ic49c0729e084d8640018bd8b1ae84158cb8f3ed1
In order to cater for distribution install tests in the roles
that run as quickly as possible, we add these.
Change-Id: I5fd21f877ee523250b718ff4e7eb49ff6a959783
Since introducing distro. installs, we have two different role
inclusions, one for distro and one for source. It seems that this
was done to send a different set of variables to each run.
With the recent build changes, all the branch variables are already
included using from openstack_services.yml so there's no need to
redefine them.
This means we can bring the two includes back to a normal role
reference, which resolves a few issues around using import_role and
brings back the ability to use tags only with this playbook.
Change-Id: I6c6651ce923530b1d2395cd387faf6e74c9b17d8
To help us get the rest of the python build simplification
work merged before we're able to remove the legacy parts,
we extend the timeout to the maximum allowed for the check
and gate pipelines in OpenStack-CI.
Change-Id: I976a7e242bcc18f77ce2ec88edcb43fb96f66c98
In https://review.openstack.org/608332 the dist name was changed
from 'congress' to 'openstack-congress'. The egg name needs to be
updated or the package will fail to install or be usable.
Change-Id: Ic9d78ca26a4a5f554724dd26c7e766e0f4b29cee
This commit adds experimental deployment of Masakari role.
It requires existing corosync/pacemaker cluster on compute nodes
for hostmonitors to operate correctly.
Corosync/pacemaker deployment stays out of OSA scope for now.
Depends-On: Ib33d7bc83f1428763f873e1155fd9e3eb4c937e4
Change-Id: Ie543885a52f013635b9f553982c3d6448e3cc3aa
In order to transition all roles to using the new venv build mechanism,
we need to make sure that the git pins override the role pins.
Change-Id: If8c8bd211e20c1f61b3ccb1237c5961843243e1e
This way the group of folks interested in keeping openSUSE Leap
working can monitor this for breakages without having to wait
for a review to be blocked on it.
Change-Id: Ica3eba3f762fe5fc2016c813c88af65ce6148c3d
openSUSE Leap 42.3 ends November 2019, so for master we don't want to
maintain the old version of openSUSE any longer.
Change-Id: I11589d8bc55d1d9320037a2ca0b294d99f7fbadd
Removed the reference to netloc_no_port
as it will be taken out from
openstack-ansible-plugins. Now using ansible 2.4's
urlsplit filter as a replacement.
Change-Id: Ib7c1a0a23ee73e355c39e4061aed05ce607cd839
Related-Bug: #1820830
Version 1.29 of openstackdocstheme includes now the build scripts
for translated documents which we have had as a PoC in that repo.
Fix Sphinx Warnings (a lot)
Change-Id: I4f9a48bd543e8d394f23491fadb6ac47f24a156a
Since change Idb9a178ca2165920804eee12a54419722aa64f51, we no longer
attempt to directly SSH to the ceph mons. Instead, we use the ansible
wait_for_connection module. One advantage to using this module is we
can pass it an ansible inventory host instead of an IP address, and
it will use the configured connection settings for that inventory
host, such as the container connection plugin.
This allows the ceph-client tasks requiring mon access to complete
without a hard dependency on an SSH connection to the mon. With this
change, we will simply pass the ceph-mon inventory hostnames to the
ceph_client role rather than a list of IP addresses.
Change-Id: Ia94ba882c2a8f568a2dcca92ee381f5833bd50af
After OSA AIO deploy octavia container and amphorae VMs are placed
in one L2 network domain but have different subnets, so it isn't possible
for octavia container to connect to amphorae VMs for configuration.
The solution is to create neutron subnet with the same CIDR as for
lbaas-mgmt, but separate pool from space where container will start
Change-Id: I477878073f8698862dd38b1249025e2f8c7a68ed
Closes-Bug: 1818441
