* Use include function instead of template.
Signed-off-by: Naseem <naseem@transit.app>
* Hardcode probe path.
There is no value in making the path configurable. By hardcoding it to /ping we remove unecessary templating and config complexity.
Signed-off-by: Naseem <naseem@transit.app>
* Refactor configmap by using implicit default values
Instead of having every default value explicitly described in the values file and put into the corresponding configmap(s), just use defaults and adjust these values as needed. This is beneficial because overtime defaults may change and it is duplicate work to match it in the values file. Furthermore documenting configurable values in the README will be much more feasible. This PR also removes the now defunct admin section of the config.
Signed-off-by: Naseem <naseem@transit.app>
* Rename config to configmap
Signed-off-by: Naseem <naseem@transit.app>
* enable api and rpc ports always
Signed-off-by: Naseem <naseem@transit.app>
* Adds support for InfluxDB Enterprise
Follows the newly introduced pattern of StatefulSets for pods, extending InfluxDB's
sts and allowing parameter for it's ReplicaSet, and adds a new sts for InfluxDB
"meta nodes" which hold distribution state for InfluxDB's high availability and
horizontal scaling.
Documentation for these new features has been added, but I can not stress enough
how important it is to follow, as Enterprise not only uses different container
images for meta containers, but also expects new container images for InfluxDB
(appropriately labeled `data` in Docker Hub) which in my experience was easy to
miss if a user is following one set of [instructions](https://docs.influxdata.com/enterprise_influxdb/v1.7/install-and-deploy/production_installation/data_node_installation/) over [another](https://hub.docker.com/_/influxdb).
A caveat of trade off of deploying Enterprise is an expectation of human
interaction when creating or scaling the cluster (`influxd-ctl` usage). This
makes Enterprise decidedly not "cloud native", but is less obnoxious when using
a StatefulSet with PVCs, and no pod auto-scaling, of which I suppose is unlikely
given InfluxData's approach to licensing.
Still forcing k8s operator involvement feels "bad", and I left in provisions to
improve this expectation in a later version, namely a headless service and using
`publishNotReadyAddresses`, such that a pod can use the Service to find
available `influxdb-meta` and `influxdb` pods to create/join an InfluxDB
cluster.
Signed-off-by: Weston Johnson <wgj@automox.com>
* Adds missing newline at end of file
Signed-off-by: Weston Johnson <wgj@automox.com>
* Add reference links for InfluxDB Enterprise
Signed-off-by: Weston Johnson <wgj@automox.com>
* Bump version
Signed-off-by: Weston Johnson <wgj@automox.com>
* Remove bad whitespace that sneaked in
Signed-off-by: Weston Johnson <wgj@automox.com>
* Hard code headless service for metadata
Metadata service is a StatefulSet, so it doesn't make sense to let users
to set the k8s service to anything but ClusterIP::None. This is true for
the StatefulSet for InfluxDB data pods, but that will be fixed in a
subsequent PR. When that happens, the Enterprise section of README.md
should be updated to remove references to ClusterIP::None.
Signed-off-by: Weston Johnson <wgj@automox.com>
* Set the default user from an existing secret
Signed-off-by: Angelo Fausti <afausti@lsst.org>
* Address reviewer comments
Signed-off-by: Angelo Fausti <afausti@lsst.org>
* update to support k8s v1.16
Signed-off-by: Jeff Billimek <jeffrey_k_billimek@homedepot.com>
* adding necessary matchLabels
Signed-off-by: Jeff Billimek <jeff@billimek.com>
* Bump major version and update readme
Signed-off-by: Jeff Billimek <jeff@billimek.com>
* bumping chart version not appversion
Signed-off-by: Jeff Billimek <jeff@billimek.com>
2. removed default resources and forcing users to add memory and cpu
requests and limits
3. bumped chart version
Signed-off-by: Deepak Kumar Sood <deepaksood619@gmail.com>
* Support list of ranges for loadBalancerSourceRanges
Signed-off-by: Wilfrido Vidana <wvidanas@gmail.com>
* Chart version bump
Signed-off-by: Wilfrido Vidana <wvidanas@gmail.com>
* Similar to other charts, this allows configurable name
overrides so 'cleaner' resource names can be achieved,
which are independent from the wanted release names.
* This is a must have if you want to integrate influxdb with
other charts, where a resource name can't be templated
and has to be fixed in the values.yaml, and you have to
deploy that in multiple instances on the same k8s cluster.
For example if you have an integration chart and
the grafana data source URI points to a fixed influxdb
service name, then without this patch,
it can't be deployed in multiple instances,
since the service name depends on the release name,
which has to be unique.
* [stable/influxdb] Add delete-policy for set-auth
Withouth any hook delete policy,
the hook resource remains in the namespace
even after a 'helm delete --purge', and
consequent installations of the chart fail with:
Error: jobs.batch "[...]influxdb-set-auth" already exists
* [stable/influxdb] Bump version
* Fix maintainer name
Matt Farina
Naseem
Jeremy Roberts
Irtiza Ali
GullomJ
wgj
spring-bu
Angelo Fausti
Dennis Wendland
Max Rosin
Felippe Mendonça
Holger Mauermann
Jeff Billimek
Jason Harris
Jorge Eduardo Villaverde
Tomasz Główka
Zach Seils
Aisuko
Ian Zink
allaVolkov
Naseem
Deepak Sood
Carlos Tadeu Panato Junior
Wilfrido Vidana
Rafael Matias
giddel
azhelev
Fulton Byrne
Kali Norby
Ferenc-
Mung