Without this patch it is not possible to have an override of
the helm values for the ingress controllers.
This is a problem, as this is inconsistent with other components
and this has reduced flexibility.
This patch solves the problem by exposing two extra overrides
for ingress: $OSH_EXTRA_HELM_ARGS_INGRESS_ceph and
$OSH_EXTRA_HELM_ARGS_INGRESS_openstack, next to the usual
$OSH_EXTRA_HELM_ARGS
Change-Id: I5b56941a6e9a585b9398099c632df349414112fa
This updates the upgrade host playbook in openstack-helm to match
the playbook used in openstack-helm-infra. The recent addition of
adding an apparmor profile to the calico chart requires us to
do the same setup on hosts in the openstack-helm jobs before
attempting to deploy calico
Change-Id: I264ba4ee8a2f24ffcbb36e28f6b91bbc114b406d
This PS updates the mount options for the nova-compute pod to mount
cgroups as read only within the pod.
Change-Id: I82e958c2865029cd4a093f62614a1e878075098a
Signed-off-by: Pete Birley <pete@port.direct>
This parameter should only be set to true for resize testing in
single-host environments [0].
[0] d64644a29d/nova/conf/compute.py (L40-L47)
Change-Id: I6ade8c90cd0916879a5addca54379ba354770c95
The update makes sure the Openstack service's cephx
user capabilities match best practices in terms of
security permissions after a site or software update.
Change-Id: I70e7f620accb186da2013ba95472777c25739cc1
This updates the single node jobs to use the recently
added minikube deployment, with the intent to reduce
overall runtime for the single node jobs
Change-Id: I6efdbf890d86bf916cef2d1a3b7eba1f6132c2f9
This commit replaces the hard-coded name for tempest jobs with the
unique Release name, allowing multiple runs of the job in the same
namespace without causing collisions.
Change-Id: I0c8aada93803875d7b0ff216c12ea67a925a510f
This begins the reordering of the openstack-helm gates. This
deprecates the single node checks/gates that deployed the entire
stack in favor of single node checks/gates that are focused on
deploying charts with only their dependencies to reduce the
number of checks/gates required for a particular change. This also
moves the armada check to experimental, and moves the multinode
checks/gates to run as periodics. This will be followed up by
additional efforts to streamline these changes and incorporate
previous work targeting the same.
Change-Id: I63b87aceefc79a7a42c325669f2b4e3abb0c961c
This will enable the TaaS GUI to the horizon dashbboard.
TaaS dashboard will need to be installed as part of the image,
else it will not try to add the panel.
Change-Id: I226d9d6e46f5b556a7baa88fcd06de8e571bcdff
- This doc ps include guide to setup multiple Ceph clusters
- One Ceph cluster for k8 storage backend (pv)
- Other Ceph cluster for Cinder and Glance storage backend.
- All dependant PS has been merged during Ceph split
Change-Id: I918eb10e7d908b1f879d9011020b6d74ac303eea
This PS removes the vmbc image from osh, and also fixes some linting
issues with the ironic gate setup script.
Depends-On: https://review.openstack.org/608689
Change-Id: I2f95445a49dfaced19cab058f94966f11c4a8877
Signed-off-by: Pete Birley <pete@port.direct>
This patch set updates the gate to by default uses network policy
for all components and enforces them in Openstack-helm.
Change-Id: I70c90b5808075797f02670f21481a4f968205325
Depends-On: I78e87ef3276e948ae4dd2eb462b4b8012251c8c8
Co-Authored-By: Mike Pham <tp6510@att.com>
Signed-off-by: Tin Lam <tin@irrational.io>
Without this patch, openstack-helm-functional and
openstack-helm-multinode are a repetition of the
openstack-helm-infra-functional and openstack-helm-infra-multinode
jobs.
This is a problem, as multiple patches are required to change a base
job, and can lead to inconsistencies.
Instead the abstract jobs should be defined in only one place, the
openstack-helm-infra repositories. This will therefore be the location
of the openstack-helm project lint, functional, and multinode "base"
jobs.
From that point, openstack-helm-infra and openstack-helm can implement
their own job variants should this be required.
This patch fixes the link dependency between the two repositories by
ensuring all the OSH jobs are using their own local job definition
("-temp" jobs) during the migration.
Change-Id: I3d3193415ba65d58c5dd9bbb1fc515f5399edece
Depends-On: https://review.openstack.org/610482
Depends-On: https://review.openstack.org/610488
Depends-On: https://review.openstack.org/611239
This enables Ceph discard/trim support with cinder volumes.
Note: that you still need to use a paravirtualized storage backend that
supports discard/trim (virtio-scsi for example).
Change-Id: I7d564ac22ee8faedda0336ee5c67f8312c9f7d06
Fix some typos that I found in documents(except specs)
* fix trivial mistakes(typos)
* fix the link error(rst->html)
* restore it's -> its
* fix k8s -> K8s
Change-Id: I6ec65e9d04441adac210cc9fd476a37a1cb9644f
This PS removes the rally ec2 credential test. As Rally left old
encrypted ec2 credentials in the database which were not being
cleaned up between upgrades and key changes.
Change-Id: Id0c5411521adebc8c977fd9a24f511f6925f16d0
Signed-off-by: Pete Birley <pete@port.direct>
This moves the job definitions into a single location, the zuul.d
folder, like it is done in the openstack-helm-infra repository.
The job playbooks and definitions of this repository will be moved
away to the openstack-helm-infra repository in a later commit.
This patch implements the first job reuse with the job
openstack-helm-lint from openstack-helm-infra's repository.
Change-Id: I233da952548d69f040771007455955131139e860
Depends-On: https://review.openstack.org/#/c/608662/
This patch set moves the default deployment to ocata from newton.
Newton zuul job is now moved into its separate job.
Change-Id: Ic534c8ee02179f23c7855d93a4707e5a2fd77354
Signed-off-by: Tin Lam <tin@irrational.io>
This adds the Elasticsearch admin password to the nagios chart
document overrides to account for the Elasticsearch plugin for
alerting on queries to Elasticsearch for particular logged events
Change-Id: I589fabf94a537730c3bfe86a6aa2af2c6fc3c88a
This moves to use a single rabbitmq deployment for the openstack
services in the armada gate to reduce the resources required for
this check to run.
Change-Id: I81bba3c9a4d9cdaad6ddcaa066441a70f1908415
If conf.nova.ironic.memcache_secret_key is not explicitly set, derive
it from endpoints.oslo_cache.auth.memcache_secret_key or use a random
value.
This means when installing charts where we explicitly set
endpoints.oslo_cache.auth.memcache_secret_key upgrades without changes
won't churn.
Change-Id: I686297e25627d88ff9bd32df0a3f7ee8afc11f58
While implementing network policies, we noticed a handful of pods created
as part of a CronJobs are missing labels causing them to be unable to
targed by the policy. This patch set adds in the missing labels found
in that effort.
Change-Id: I1ca3cfd68ff20dc39a1e952414f3dddd3fc8d3b4
