It's a server project and so needs to explicitly list these jobs
(which most non-server projects get from the publish-to-pypi
template).
Change-Id: I9be5eb1f7ad0264ce91a677ecda99d5cc0c493a1
- allow os-ansible-deployment core group to approve reviews for master, stable/* and proposed/* branches
- allow os-ansible-deployment-release to create tags and new branches
Change-Id: If1ada9841d5de4a0f584c9e765056d273a579932
This patch inserts an all-in-one server deployment as a non-voting check. It
will later be promoted to a gate once it's confirmed to be working as required.
Change-Id: I310bdf13e7c62c8f81d7a0effb8a7b53f683b36a
In order to ensure that the syntax and lint checks are only processed on
master and do not fail on older branches, the check is only being set to
run on master.
Change-Id: Iea9c2dd5e6fd112775e0fd073b93c0e842af7f27
Since there are many lines of python 2.6 deprecation, add a comment at
the end so that it further jobs can be added below the list and it's
clear that they are not part of the 2.6 deprecation.
Add comment for python26-jobs about the deprecation.
Change-Id: I02007ba9eee8f6458446b9fb292647804a6c7543
Castellan is a Key Manager interface for OpenStack projects that do not
want to integrate with Barbican directly. Barbican will provide an
implementation of this interface, but projects can choose to implement
the interface with an alternative Key Management System if Barbican does
not meet their requirements.
A CR has laso been sent to the governance repo to add Castellan to the
Barbican program:
https://review.openstack.org/#/c/138875/
Change-Id: I076e04e5167e2ff00b94142e86c242de36e766da
Fix the python interpreter used by layout-checks.py so that the "tox -e
zuul" target works correctly on a local system.
Add the PyYAML dependency needed by layout-checks.py to the tox
configuration for the zuul target.
Add the zuul target to the default list for tox.
Change-Id: Ic4db9fc3bb236d36365abfe966eb9e7769896011
When 'tox -e pep8' fails we don't end up printing
the pip freeze output since pipefail just bails out.
We should capture pip freeze when flake8 passes
or fails. The pipe seems to be a vestige from a previous
attempt to save the output into a separate file, we
dont need that anymore. Save the exit code from
the tox command and use that when we exit at the end
of the script.
Change-Id: Ic4590b7a8260a4a652608b1aa8d02beb786143ea
The projects that have their ACLs modified here have had third party CI
vote on them in the last 3 months. Add project specific CI group ACLs so
that these projects can directly manage which accounts are allowed to
vote on them.
Change-Id: Iaf9cbb86cd3077f28bf3c5f02de4718314ac1eef
github repo name is stackforge/os-ansible-deployment
launchpad project is openstack-ansible
Added the groups mapping to ensure correct functionality
Change-Id: Ic0a310510814829577a1c27c984df536b4a2d295
This change adds the gerrit irc bot to the openstack-ansible channel.
The bot has been setup to watch the branches **stable/icehouse**,
**stable/juno**, and **proposed/juno**.
Change-Id: I00a373744faeef9a8243c9df5edd8c83775f3cd8
Bandersnatch and pypi have a failure case where uploading two files to
pypi very closely to each other results in bandersnatch thinking it has
synced serial X+1 which includes both the sdist and wheel but it really
has only synced serial X. But since X+1 is what bandersnatch thinks it
has done it will never sync the actual contents of the X+1 serial (the
second file uploaded).
To make matters worse the index.html for the project is synced assuming
both artifacts at serial X+1 were synced. This means any fetches for the
second artifact fail.
We can work around this by always uploading the wheel first. Pip prefers
wheels to sdists so uploading the wheel first ensures that the wheel is
synced properly, pip will prefer it, and our jobs should continue
running.
Eventually upstream pypi and bandersnatch will need to fix this bugs
properly but this is a reasonable thing we can do locally.
Change-Id: I26062b9fefa5ce484b63e2d0ee02e0ecbe64555f
Fixed ips do not work with current nova-cells. Floating ips partially
work depending on which database the floating ip pool(s) are created in,
but they don't just work out of the box. So these tests should be
disabled so we can get closer to a passing job.
Change-Id: I77c011ed3a36757519e07030ced779cee8b108b9
Refactor the gate-designate-devstack-dsvm job to be a reusable
job template, turning it into gate-designate-dsvm-powerdns, and
create a second instance for the BIND9 designate driver.
Change-Id: I7c5321e82caf8d54cf74095cea18dd1ea5b5bac5