A recent change in keystone [1] has deprecated the token auth
mechanism that we used. We reintroduce it temporarily while a more
permanant solution is worked on.
[1] 5286b4a297
Change-Id: I4d585733a9abd201c1b0680e6196dd2a36db3c7e
Closes-Bug: #1545292
The CentOS packages install a default version of the paste file
but don't install them in the /etc directory. Kolla doesn't
configure the paste files, so this file never gets setup properly.
With the recent change in Keystone around the default admin_token
being changed, this triggered Keystone to return a 401 Auth error.
A follow-on patch from Sam edits this file, and that patch breaks
the build without this dependent patch.
TrivialFix
Change-Id: Ib1568e186bdd7d19b7e5af151388197755902488
original extend_start.sh try to write temp files to a non-existent folder (/home/horizon) and cause horizon failed to start. It is fixed by moving the file to /tmp.
Partial-Bug: 1543429
Change-Id: I8d8778be755a4db9e57d6ff7aaee623547c4243e
This prevents failure when the directory already exists.
This commit fixes a failure in centos-binary because of a change in the
RDO packaging, where the required directory is now created for us.
Change-Id: Idd3e15802c3e3fd363e1295111ec12948d566781
Closes-Bug: #1543417
The python-cinder package now pulls the required dependency on
python-oslo-policy:
a22f32c619/openstack-cinder.spec (L143)
TrivialFix
Change-Id: I504546e75b5a237fc4aac89bd9a5c34f09570817
The python-nova package now pulls the required dependency on
python-oslo-policy:
036149e93f/openstack-nova.spec (L389)
TrivialFix
Change-Id: I472f600710cb743ef6ede01e722a79b1451cf295
The python-neutron package now pulls the required dependency on
python-oslo-reports:
774036f7f9/openstack-neutron.spec (L125)
TrivialFix
Change-Id: I6cb4d0d9601c910cf631dddcf715a7f4fc9754e3
*** Requires Docker 1.10 which is released ***
Documentation will be in the next patch. You must set the following
in your docker.service daemon control file for propogation to work:
[Service]
MountFlags=shared
======================================================================
Thanks to mount propagation in Docker 1.10 we can use thin containers
finally! This is extremely useful to operators since now they can
access the network namespaces from the hosts (outside the neutron
container). But additionally it allows us to implement the VPN agent
and other services easier.
Neutron containers and the neutron role are brought into the standards
of the new Kolla. Completely with drop-root and ansible formating
updates.
The ip_wrapper.py script was (thankfully) not needed so it has been
removed from the repo.
Partially-Implements: blueprint upgrade-neutron
Change-Id: Iaf5555283240457e1912459f397a6393d886fba1
Because we use rootwrap in a venv we need to update the exec_dirs.
Without doing this *some* commands will break for various reasons that
I won't get into in this review in detail. But this is required for
neutron-l3-agent in a container + drop-root.
Change-Id: I1a09f7188fdd501b7ce251d9f2fb0e5b10222142
Partially-Implements: blueprint drop-root
The install type is converted in kolla-build, so it will never
fail in dockerfile, move the check to kolla-build just above
the install type converting.
TrivialFix
Co-Authored-By: Jeffrey Zhang <jeffrey.zhang@99cloud.net>
Change-Id: I1500d3b47e909f94ea9f68c5245297733f63a70b
This change is needed for clarity. We have a kolla-ansible script.
We have a kolla-mesos repo. We plan to have a kolla-ansible repo.
Already we have had far too much confusion about whether we are
talking about the container or the project. Naming this kolla-toolbox
eliminates all of that confusion and its probably a bit more accurate
of a name too.
Closes-Bug: #1541053
Change-Id: I8fd1f49d5a22b36ede5b10f46b9fe02ddda9007e
Current Swift playbook is based on the preassumption of AIO setup.
However, if one goes with default multinode setup
(ansible/inventory/multinode), it follows the P + ACO deployment model,
which proxy-server runs on controller nodes where ACO
(account/container/object services) run on storage nodes.
It will break because swift proxy-server no longer has access (it
shouldn't have) to /srv/node path. This change ensure disk mounting part
only happens on storage node. It also moves chown from proxy-server
Dockerfile to rsyncd because no matter with PACO, P+ACO or P+A+C+O
model, rsyncd is always running on each storage node.
Change-Id: I3aa20454902caa9c84d3901bb91e4e4c93ac5f34
Partially-Implements: blueprint swift-physical-disk
Closes-Bug: #1537544
Ansible installs pytz from pip which is newer then then pytz in
RDO repositories. As part of this process, it removes pytz
which corrupts the python installation in some way. This causes
further pip usage to produce a backtrace causing the kolla-toolbox
container not to build.
Mixing pip and RPM is not good.
Change-Id: I413468641bad5476cfcba61cb0e2189c31f3d682
Closes-Bug: #1545215
Steve is tired of maintaining a copr for Magnum. People bug him
all the time to update the rpm for RDO. The RDO community
has offered to take on the maintainance of the Magnum RPM. As this
RPM won't be in current-passed-ci for some time so it needs
to be pulled from current repo for the foreseeable future,
possibly nearing the release of Mitaka.
Change-Id: I9cfb02ab828251ef5bf40ca236f18b5f0f715e34
Closes-Bug: #1539325
There is no reason to have a hostname-unique pidfile in the container
as we currently have. This posed problems with kolla-mesos reusing
the same script. Since there is no reason for this pidfile to be
configurable in path _at_ _all_, we hardcode the path.
Additionally, we adjust the file perm change to only update the perms
on the folder if it is not already properly set.
This also incorperates a kolla-ansible file in the bootstrap process
which follows our other container techniques of using the idempotent
creation of a volume in the bootstrap process (see nova)
TrivialFix
Related-Bug: #1538136
Change-Id: I2380529fc7146a9603145cdc31e649cb8841f7dd
$(hostname) is Ceph Monitor name in extend_start.sh,
{{ ansibe_hostanme }} is Ceph Monitor name in ceph.conf.
$(hostname) not always equal to ansible_hostname , that
makes ceph_mon container can't start.
Closes-Bug: #1538870
Change-Id: I312bf8d74c855aa4c72f12285e3092df96f60048
Currently the only consumer of ansible find_disks module is Ceph. And
Ceph OSD deployment in kolla uses GPT partition label to detect and
identify disks for Ceph OSD use. This is not always true for all the
deployment.
The change here extended the find_disks module by:
- adding `name` argument to find disk by either partition name or
filesystem label matching
- `partition_name` argument now becomes an alias to `name`
- adding `match_mode` argument to allow prefix matching. It is used for
swift disk detection.
- return `fs_label` key / value in result for disk mounting purpose
Change-Id: I9c93400c1826f5148acf09e9fbe555e358dfdfcc
Partially-Implements: blueprint swift-physical-disk
