This PS includes the release name in the cluster role to prevent
colision if the chart is deployed multiple times in the same
cluster.
Change-Id: I7166e5ee25b3d4c89879393c5f84c869585a2681
This removes the etcd dashboard as a default dashboard for grafana
as it currently doesn't work out of the box
Change-Id: Icd435268756336113219df66068f62c8870ed9f9
This adds templates for a prometheus exporter for fluentd to
adequately capture log metrics and fluentd service metrics for
consumption by prometheus
Change-Id: I6d6a8c2be07af819dc6d99b8ce5f1d4b635a69f0
Adds "helm-toolkit.utils.merge" which is a replacement for the
upstream sprig "merge" function which didn't quite do what we
wanted, specifically it didn't merge slices, it just overrode
one with the other. This PS also updates existing callsites
of the sprig merge with "helm-toolkit.utils.merge".
Change-Id: I456349558d4cf941d1bcb07fc76d0688b0a10782
This PS removes the duplicate labeling of K8s namespaces everytime
the kubeadm image is run - as this action is performed by the
container itself while deploying k8s. It also updates the playbook
to label the kube-public ns.
Change-Id: Icfaabe54d263e0bbc52f2ac11835cb5d92ad32a4
This adds a chart for an elasticsearch exporter to expose metrics
for prometheus. This also moves the exporter to be included as
part of the chart it's meant to target as opposed to its own
separate chart
Change-Id: I491f4d1efba633827d8a6255218daeb9d427f922
This updates the prometheus metrics gathering role to dynamically
detect deployed namespaces instead of using hardcoded tasks for
each namespace. This allows the use of this job in openstack-helm
without needing to manually add additional namespaces
Change-Id: I2304aceab8dcda5471af708f57924193f03ee8f3
This patch set fixes the namespace labels and adds in a new log file to
describe all namespaces to troubleshoot gate issues where labels are not
applied properly.
Change-Id: I34c43c345f9a49df8d7fcf2e7824220a22698d46
Signed-off-by: Tin Lam <tin@irrational.io>
Zuul no longer requires the project-name for in-repo configuration.
Omitting it makes forking or renaming projects easier.
Change-Id: I35277bfa2de254337854c6058e32f91ac1d621c8
The update to helm 2.8.0 resulted in issues with releases not
registering as ready before timing out. This rolls the version back
until those issues are addressed
Change-Id: Id58f85f150054d82c047bd360258ebe9e571360f
This adds node selectors to all templates in the elasticsearch
chart, as they were previously missed
Change-Id: I34ea5751663b2e993c5f73a78a1f91133919752c
The volume mount name for the permissions init container was
overlooked when the PVC for alertmanager was removed and the
volume renamed. This changes the mount appropriately
Change-Id: I5db6594a3192ec78354e5f3d3d41e96317488664
The openstack-exporter doesn't have helm tests currently, so there
is no need for a helm_tests image key.
Change-Id: I7d1f737170a343e4ec66c4d7ace3ba76cdffb329
* Ingests the bird templates so that we can override them
to support things such as custom BGP ports (listen) and
neighbors (remote)
* Supports announcing addresses that are within the
.Values.networking.bgp.ipv4|6.additional_cidrs list
in support of ingress controllers that can create
dummy interfaces and assign addresses to be announced
* Introduces a new job to perform calicoctl manipulation to
support manipulating the mesh, adding peers, and changing the
ipPool settings which is value driven
* Support custom port binding and specific interface binding
to allow custom BGP port selection for IPv4 and IPv6
* Instantiates calicoctl as a utility on hosts
* Adds a new function to helm-toolkit to retrieve the http
or https prefix for an endpoint
* Supports https based etcd backends with new certificate
parameters
* Finally, introduces more strict bgp listening to allow
multiple hostNet bgp speakers to run in parallel
Change-Id: Ib4d00befddbd8498b9dcc693409b8b2577458497
This removes the pvc in Alertmanager and changes the default
storage_class to readwriteonce. Now that Alertmanager uses peer
meshing, it's not required for the replicas to share a common
volume claim
Change-Id: I24290264cb0e552a143a56faa753289f073c47b9
This disables the Elasticsearch snapshot repository backed by
NFS by default as the curator job for snapshots is disabled by
default, and should make no assumption that NFS is deployed by
default
Change-Id: Idc74cfb80fcb4c4741d82c6d0ce63fd90a8c919f
This increases the default scrape and evaluation intervals to match
those of the prometheus service upstream with the aim to reduce
prometheus's resource consumption. It also adds configuration
parameters for the min and max block durations that series can
span
Change-Id: I7f9352413a273fbf680b892ba26e30cf27bae232
This PS uses volumeClaimTemplates to provide multiple
pvcs and pvs to prometheus statefulset's multiple pods.
This gives ability to provide read-write-many access
backend by a read-write-once storage class.
Change-Id: I53d1b866c0c87f0833941b612d3acfbe5742744f
This PS is kube-state-metrics for kubernetes version 1.8.
Using kube-state-metrics:v1.2.0 image makes kube-state-
metric pod work properly. Also, gives authority to list
endpoints, persistentvolumes, and horizontalpodautoscalers
by adding them to clusterrole.
Change-Id: I705b29c321b0162740744afa8573dc6ae75bcc60
There was a change in the upstream reference httpd image for
apache that changed how modules were built for apache.
This change adds the required fix to accomodate the change.
See isssue here https://github.com/docker-library/httpd/pull/87
The Elasticsearch image tag was updated to accomodate the kernel
versions used in the gate as part of the kernel update playbook
See https://github.com/elastic/elasticsearch/issues/28349#issuecomment-360233779
The openstack-exporter binary was changed to reflect changes made
to the openstack-exporter image
Change-Id: I1deb9e7cde794421dd33fade566c2a9fdb5007e6
