|Age||Commit message (Collapse)||Author||Files||Lines|
Now that we got a promotion, there is a container image for fluentd in
tripleomaster registry. We can finally re-enable the containerized
fluentd service in scenario0001.
(cherry picked from commit c795e748d2e0fe1299919872250d6951246c1365)
Mixing containers and BM is currently not working. Once the master
promotion will take place we will have a fluentd container and
can readd fluentd as a container and the problem should not re-occurr.
(cherry picked from commit 35d91ddc6d11bb2696321fff4593d5cca3b0cba8)
This commit brings the multinode containers scenario files closer to
their BM variants to add missing services and turning pacemaker on.
These require refactorings in OOOQ in order to support non-containerized
to containerized upgrade jobs across releases. Ceph-ansible is also
going to be switched separately.
Co-Authored-By: Jiri Stransky <email@example.com>
(cherry picked from commit c504f83c28b986ceb2b92cc0077959158bd11df7)
Use ceph-ansible to match the non-containerized variant of
Co-Authored-By: Martin André <firstname.lastname@example.org>
Co-Authored-By: Giulio Fidente <email@example.com>
(cherry picked from commit 4e85813d368b94f588471d9bdc4a04d04d3be541)
We have disabled mongo by default in containers via:
Id2e6550fb7c319fc52469644ea022cf35757e0ce Disable mongodb by default
Ie09ce2a52128eef157e4d768c1c4776fc49f2324 Containerized mongodb, disable by default, fix upgrade
Let's not use it in scenario002 either.
NB: Not entirely clean cherry-pick due to scenario002-multinode-containers.yaml
having many more services in master than in pike.
(cherry picked from commit 900416d9809bf4446c0c037128edb033ab9b3bcc)
(cherry picked from commit 7e398bf18910e062415ce4e70236ce98577aed13)
We should not pass any hardcoded value for monitor_interface and
rely on monitor_address_block only instead.
Also removes journal_collocation which is not consumed by
newer (and stable) builds of ceph-ansible.
(cherry picked from commit 3e90ae3df5a7c5491672254733ceac163b34a395)
(cherry picked from commit 06331a830e8923a9dc2ef8c15f2f1bf9d1d58ba1)
(cherry picked from commit f37fe4f903f429b43d22b485c29547f576ec7269)
As per Ceph docs  we should default pg_num and pgp_num to 128 when
using less than 5 OSDs.
This same change was applied to the ceph-ansible profiles with .
Also updates the CI environment files to continue using 32 where we
deploy a single OSD.
(cherry picked from commit e17ae7620e03790da0d29092ab42e8089b2e8d11)
node_private file doesn't exist anymore, use sub_nodes_private
(cherry picked from commit 3e8de70bd5a8c43389432d484189d4de5fc0ae2f)
This was needed to make the upgrade job on Ocata->Pike passing, and we
now need to remove this to improve the argument order in OOOQ for
deployments with scenarios.
This shouldn't be backported to Ocata (at least not before we make the
split between deploy scenario and upgrade scenario).
Co-Authored-By: Martin André <firstname.lastname@example.org>
(cherry picked from commit 31550b42027588d82f01db6956c1efaf02d58558)
This commit brings the scenario004 file closer to its BM pendant. We
need to start with this one to address a chicken-and-egg issue with
Co-Authored-By: Martin André <email@example.com>
(cherry picked from commit b4d0a81e55ad51ecdaf2e923f794418ac77cfc57)
We need to reuse the ceph_conf_overrides structure provided by
ceph-ansible for both user provided configs and TripleO managed
configs. This change merges the special user facing parameter
with the TripleO generated configs.
Also adds osd_scenario and osd_objectstore params for compatibility
with newer ceph-ansible versions.
(cherry picked from commit 32bc2abf14af4ca1449e18b848e2be3cff013987)
The containerized mongodb service was removed from the default service
list from 'environments/docker.yaml' in
Ie09ce2a52128eef157e4d768c1c4776fc49f2324. This commit re-enable the
containerized mongodb in scenario002-multinode-containers.
(cherry picked from commit 13d2bdf41ca3b726acff0b94dae6d394dab4af23)
We do not want a default value for the container image name parameters
and expect deployers to set this value instead.
(cherry picked from commit fcc3259891ee67956d63c37217acdb999bc4bb65)
See full context on https://bugs.launchpad.net/bugs/1713612
but this service isn't containerized yet, so the plan is:
- in Pike, we'll run scenario004 (baremetal) and test bgp-vpn and l2gw
- in Queens, we'll run scenario004 (baremetal at the beginning) but
scenario004-container will be the default and we'll re-add the 2
services when containerized.
(cherry picked from commit fde4ff2c64f374e109dbb7da87cc7d72da5e0ef5)
(cherry picked from commit 02cd34d148d6abf11cc64852f7931cbd4bccf767)
This service is necessary when we containerized TripleO with
The service is added also to non-containerized scenario lists, because
the aim is to get rid of the -containers.yaml variants eventually.
This shouldn't affect any jobs that don't include docker-ha.yaml. The
resource registry entry is mapped to OS::Heat::None by default, and
docker-ha.yaml maps it to actual containerized clustercheck.
(cherry picked from commit 5b805cb37eec3097552314c6ce43c02c2a604d81)
The containerized implementation of tacker is incomplete in THT,
and relies on the pre-pike single "tacker" container. Container
builds using the final pike release of kolla build three tacker
containers to have seperate conductor and server containers.
According to this bug, tacker does not even work without this
conductor. Our scenario job needs to be updated to actually test
tacker is working.
This will need to be backported to pike, and we can work on
better supporting tacker in containers in queens.
- Set gnocchi archivepolicy in scenario001 job to high
- Set polling interval to 15 seconds instead of 300
(cherry picked from commit 0855d4c7b12d27721044ab09ca0d6e8f188d2e90)
This patch adds the support to containerize OVN services for the
OVN db servers do not support active-active mode yet. It does support
master-slave mode supported through pacemaker, which will be supported
in a later patch.
Presently the tripleo container framework doesn't allow to start a
container in only controller 0 (or bootstrap node). OVN db servers and
ovn-northd are started on all the controllers, but only the OVN db
servers running in the boot strap controller are configured to listen
on the tcp ports 6641 and 6642. OVN neutron mechanism driver
and ovn-controller's use the ovn_dbs_vip to connect to the OVN db servers.
Haproxy configures all the controllers as back ends, but only OVN db
servers running on controller 0 respond since only they are configured
The OVN containers running on other controller nodes do not interact
any way, but are wasteful resources.
This patch also adds the scenario007-multinode-containers CI template.
(cherry picked from commit e7cd03d2f0fcd8e3069246ced94f1a83869b8bea)
This service isn't being started by default and it's required in order
to ceilometer collect data about cpu, which is required to run tempest
(cherry picked from commit a2ce6d238cd7e72029fdd6e6364a0501f3bfc835)
This containerises Barbican API in TripleO
(cherry picked from commit 6d338b809accea4d3ba09ca8363b1a97ed79b658)
EPEL is required by the Ceph RPM repo maintained at ceph.com but
not by the CentOS SIG repo. The other packages should be already
installed in the infra image.
This change overrides the list of required packages in ceph-ansible
so that it does not install epel-release nor additional unneeded
This is required for t-h-t to generate the appropriate hieradata.
Add a docker service template to provide containerized services
logs rotation with a crond job.
Add OS::TripleO::Services::LogrotateCrond to CI multinode-containers
and to all environments among with generic services like Ntp or Kernel.
Set it to OS::Heat::None for non containerized environments and
only enable it to the environments/docker.yaml.
Signed-off-by: Bogdan Dobrelya <firstname.lastname@example.org>
Swift is already deployed on scenario002, and we want to keep
basic multinode as basic as possible with only the minimum so it runs
faster and we can use it for early tests in our CI.
to use ceph-ansible instead of puppet-ceph.
It doesn't work yet, see:
... until https://review.openstack.org/#/c/474327 is merged.
In the meantime, let's test the scenario with Barbican like before.
This change renders the network IP maps and hostname maps for
all networks defined in network_data.yaml. This should make it
possible to create custom networks that will be rendered for
all applicable roles.
Note that at this time all networks will be rendered whether
they are enabled or not. All networks will be present in all
roles, but ports will be associated with noop.yaml in roles
that do not use the network. This is in accordance with
previous behavior, although we may wish to change this in
the future to limit the size of the role definitions and
reduce the number of placeholder resources in deployments
with many networks.
Note that this patch is a replacement for original patch
https://review.openstack.org/#/c/486280, which I was having
trouble rebasing to current.
Partially-implements: blueprint composable-networks
When the OSD pool size is unset it defaults to 3, while we only
have a single OSD in CI so the pools are created but not writable.
We did set the default pool size to 1 in the non-containerized
scenarios but apparently missed it in the containerized version.
We need to test gnocchi with swift backend. So adding
telemetry to scenario002 job to cover that.
Presently the ovn-controller service (puppet/services/neutron-compute-plugin-ovn.yaml)
is started only on compute nodes. But for the cases where the controller nodes
provide the north/south traffic, we need ovn-controller service runninng in controller
nodes as well.
- Renames the neutron-compute-plugin-ovn.yaml to ovn-controller.yaml which makes more
sense and sets the service name as 'ovn-controller'.
- Adds the service 'ovn-controller' to Controller and Compute roles.
- Adds the missing 'upgrade_tasks' section in ovn-dbs.yaml and ovn-controller.yaml