This updates the prometheus rule for checking for terminated
containers in pods. The previous rule checked for any terminations,
which raised alarms due to completed containers in jobs
being included, which isn't desired behavior. This changes the
expression to check for any containers that have terminated with
a status other than completed
Change-Id: I88e533a56f81f81bd1a81420ecfb7d43ac9e2d0b
This moves the ldap chart to openstack-helm-infra from
openstack-helm, allowing for ldap to provide an authentication
mechanism for components of the LMA stack, and can still be used for
keystone in openstack-helm, as openstack-helm-infra is a required
project
Change-Id: I211bc47c7a3ae875614102c8f64daa1099f702e8
disable the local nameserver as it interferes with the k8s
dns-service and other local resolvers used for development use.
Change-Id: I32c8e752675b037610f77c8baee1c1636d5a032e
Signed-off-by: Pete Birley <pete@port.direct>
This PS places the role default vars within the appropriate role,
in addition it exposes kubeadm's selfhosted param and deploys the
kubeadm binary to master nodes.
This PS exposes the selfhosted param, and deploys kubeadm to master
nodes.
Change-Id: I5ad1b593a711ffe353b012394d54044dede0691d
This PS stops pulling the charts default images, as the make file
target cannot target over-rides - resulting in longer gate runs with
twice as many images pulled than required in some cases.
Change-Id: I04c1d43d6ac9d2f509604709139583b085b406e3
See: https://review.openstack.org/#/c/566738/
This PS moves node-exporter to use the endpoints section and
lookups to set the ports it serves on.
Change-Id: Id6d71b12e531375e792384ac4410bce74170d033
This moves to using bash scripts for deploying the charts for the
multinode gates in osh-infra as opposed to using the ansible
chart builder
Change-Id: I2951ccf57250a5e6e272872f2d6f7a81bd11f184
This ps removes the namespace selector for discovering alertmanager
instances, as it's not required
Change-Id: Ie4dc40f761096d497293d6d98b2bbb906d382101
Move to v0.3.1 of kubernetes-entrypoint which has 2 breaking changes to
pod dependencies, and also adds support for depending on jobs via
labels.
Change-Id: I2bafc2153ddd46b3833b253a2e7950bccbccf8ed
This updates the htk-manifests with the updates required to use
the updated dependency checking for dynamic dependencies
Change-Id: I696459ca8b29a80a3096a9d33c1a68b1543534c9
Updates the service discovery mechanism used by Prometheus to
identify Alertmanager instances to push alerts to. It moves to
use the 'application' label to identify Alertmanager pods instead of
searching for pods by the label 'name', as the previous definition
was resulting in empty results for Alertmanager targets
This also fixes the name of the prometheus label used to track
alerts for kube-controller-manager, as it was defined incorrect
previously
Change-Id: I1fb194550baf803435722e3a01892e49b44259d1
Curator is unable to use environment variables for configuration
values if the configured option contains more than the env
variable. In the case of the http_auth value (which expects
user:password), using ${USER}:${PASS} prevents curator from
successfully authenticating to elasticsearch. This moves to
dynamically define these values in the configmap if the value is
empty
This also updates values for curators actions to target logstash-
indices for its actions
Change-Id: Id5b49171e00847432e4ab0cf4be60005b70c21e3
This ps includes the following grafana dashboard changes:
- Renames the OpenStack dashboard title
- Removes redundant kubernetes dashboards
- Fixes datasource for the nginx dashboard
- Fixes templating variable for rabbitmq dashboard
Change-Id: I2fa1ff606746ce1f51d2ed01788bb5282bd53dfc
This ps proposes adding a common template for the image_repo_sync
jobs for consumption by the charts
Change-Id: I48476d1e4fd94bd1b08b13b46983e3d999f8d8ca
This adds the local image registry endpoint to elasticsearch,
fluent-logging and grafana. This endpoint was missing from the
values.yaml in those charts
Change-Id: I30dc1f0cab40ccf8a493e13f407e2f0d37af1eee
