We adjust the defaults and allow the same variable names to be
used for overrides. Neutron defaults to using .1 in a subnet
for the gateway, so we don't need to set it. Neutron will also
manage the allocation pool by default, so we don't need to
set those either.
Change-Id: Id7bce7072292423f2e2dac290e16e83ae7a94693
If the facts for the hosts are stale, or not present,
the playbook fails unless facts are gathered from the
container hosts. They are required because the "Ensuring
containers creation, connection and good behavior" play
uses host vars from the hosts.
Change-Id: Id4daf24f10e61a1800d73093c2aa5b9724a1c615
1. If the load balancer host is not the same as the control
plane hosts, then 'Ensuring haproxy runs' fails unless
facts are gathered.
2. The value for 'repo_release_path' is set in the vars file
'defaults/source_install.yml', so the repo check fails
without reference to that vars file.
3. When running 'Sanity checks for all containers' not all
host and container facts are gathered, so the play fails
due to missing facts unless facts are set to gather.
4. Instead of verifying both localhost and the utility
containers for being ready to run the os_* modules, we
only need to verify whichever one is the designated
'openstack_service_setup_host'.
5. The memcache server connectivity test should have a short
timeout, otherwise the task hangs for ages if it isn't
working.
6. The rabbitmq vhost name is corrected and set consistently,
otherwise those tests do not work.
7. The rabbitmq test venv fails to build for two reasons. One
is that the venv creation fails, because virtualenv tries
to download the latest pip/setuptools and fails due to the
pip.conf restricting it. The second is that the python
package 'pika' is not on the repo server. The task is changed
to make use of the common python_venv_build role and to make
use of pypi as an index when building the wheel.
Change-Id: I6f5f4a1bd55abc78ad5993076719a3ac5914af1d
In https://review.openstack.org/587586 we removed everything
except the playbook. This closes that loop.
Change-Id: Iea81197e10d6e85b723b719ca704421a7f45d927
It also seems that Ansible became more strict about the permissions of
the current working directory[1] so we need to either move our ansible.cfg
to a better location, or get rid of it. In the zuul test environment, the
location is world-writable.
Given that this is an ansible.cfg used for tests only, that its content is
(mostly) duplicated in the wrapper's RC file and that implementation of
avoiding the wrapper was done when we had vars plugins (which we no longer
have), I think it's safe to remove it and use the wrapper instead.
Using the wrapper ensures that everything is set in one place, so it's
easier to find. In the AIO bootstrap we then just make sure we override
things that we do not want to interfere with the AIO bootstrap - then unset
them later so that the defaults get applied from then on.
To also reduce the duplication of the bootstrap-aio.sh script in the
gate-check-commit script, we source bootstrap-aio from gate-check-commit.
[1]: ff980afefd
Change-Id: I6a673c3c4f228d11a9d93766021409a5b1672f63
When force_master == true, the code did not try to get tag information
from external roles but it tried to get the SHA for the specified
$osa_branch. External roles do not have stable/* branches, so the result
was an empty version. We fix that by simply falling back to master
branch if 'force_master' is set to True.
Depends-On: https://review.openstack.org/#/c/598246/
Change-Id: I73b8591217298511132d67f7a83ef27b093c02e4
The upstream PR[1] has been merged so we can switch back to using the
master branch for ceph-ansible
[1] https://github.com/ceph/ceph-ansible/pull/3015
Change-Id: Ibca788d304e67cd5a73283a9bc420d1772139b78
By default glance has ~10GB available to cache images on within it's
container. Some images exceed that size causing image uploads to fail.
To work around this we can mount the image cache directory on the host.
This has the same danger as bind mounting the image directory (filling
up the host disk), so it's not a new danger, just one to be called out.
Change-Id: I887180b165a802a8d3635c2ad5b2284a123e230d
This also adds a git track branch for tempest master, so that
tempest gets bumped by the releasing tooling, like the other
files [1].
[1]: https://review.openstack.org/#/c/596391/
Change-Id: I5d20ec377cc610090bcda75593e16093b7046c4a
The VNC path has been /vnc.html which means that no backends will
ever go up with the default option.
Change-Id: I7065fb3aa627c80655f0e1cf01742a6115a09678
This variable is not defined, causing implementations
of congress in the integrated build to fail.
Change-Id: Iaf2880866d1cc3780fec47fdf429c64227db914f
Adding this example conf file would guide user to set correct host
name in AIO case. An experimental test for barbican has been added
which uses the barbican tempest plugin.
Depends-On: https://review.openstack.org/596408
Depends-On: https://review.openstack.org/596541
Change-Id: I10a07d5e4b2b3fbf122a79f15e27a9ca34799221
There is no reason to rely on comments in the files
used to bump the SHAs of upstream repositories.
Without this, a contributor can easily add a comment
and break the tooling, which relies on regexps and
well formated files to behave.
Instead, as this is a YAML file, we should rely on
a YAML parser in the release tooling, for more
robustness.
This fixes the issues by adding an extra key
(_git_track_branch) per repository, showing what to track.
This could be used in release tooling, and is not hurting
the existing code, as the extra keys are not loaded:
Those files are only consumed by the py_pkgs lookup, which
won't read this extra key.
Change-Id: I8fe08f389160e7be17dd56afde22173778540855
Currently the role defines the repo, but there is no SHA pin
in the integrated repo, so the repo always builds from the
head of master. This will cause inconsistency with deployments
but, more important, will eventually fail to build due to
inconsistent requirements with the requirements repo SHA pin.
Also, due to their being no group assigned in the role, this
repo and its wheels are built regardless of whether neutron
is, which causes build failures.
Change-Id: Ie27d10026600ae189790c03a8d380dc0b09131d2
The patch does a few things to clean up our job management:
- Refactor all jobs into project-templates in order to very
clearly identify which jobs are voting and non-voting.
- Drop all `voting: false` stanza in jobs, the infrastructure
team suggests that we place those in job assignment, not
in definition.
- Set CentOS 7 job to voting for metal, as it is finally passing
now.
Change-Id: Iaad7a3b50e14d5f0ef25c52bcc070b7601d46a8f
The verbose output is far too verbose, making it hard to
find the results of the test in the console log.
Change-Id: I540aba6beb536ac589100ee9374cc2a157fa621d
Updates to the container common task file were made to correct an issue
which would result in a new container not getting restarted when a bind
mount was added.
Change-Id: I29948d54a015e367fbdef4d10360c1a5f544b550
Signed-off-by: Kevin Carter <kevin.carter@rackspace.com>
On bionic the network bridges refuse to be up, because the underlying
bond interfaces do not come up.
This change removes the bonds which were really only decorative, and plugs
a dedicated dummy interface into each bridge. This simplifies the config
and the bridges should come up as soon as the dummy interface is attached.
Change-Id: I2662df94929c30190cd5d303588fdc9a99370e87
This is a mechanically generated patch to switch the documentation
jobs to use the new PTI versions of the jobs as part of the
python3-first goal.
See the python3-first goal document for details:
https://governance.openstack.org/tc/goals/stein/python3-first.html
Change-Id: I7388eb65c2925c497d0fce808cb439c2970648b4
Story: #2002586
Task: #24319
This is a mechanically generated patch to complete step 1 of moving
the zuul job settings out of project-config and into each project
repository.
Because there will be a separate patch on each branch, the branch
specifiers for branch-specific jobs have been removed.
Because this patch is generated by a script, there may be some
cosmetic changes to the layout of the YAML file(s) as the contents are
normalized.
See the python3-first goal document for details:
https://governance.openstack.org/tc/goals/stein/python3-first.html
Change-Id: Ic8ca6ccdfa0b6706ef3dbd49284482a741e78fa2
Story: #2002586
Task: #24319