grafana-operator
grafana-operator copied to clipboard
grafana
Unified Alerting - implement a CRD for Grafana 8 Alert Rules, Alert Groups, Contact Points, and Notification Policies
Is your feature request related to a problem? Please describe. No existing problem.
Describe the solution you'd like Unified Alerting is a great new feature, but the current grafana-controller does not support a CRD definition for the decoupled alerts, contact points (like a Notification Channel), and notification policies. The desired solution would be CRDs for each of these so that the new alerting tools are equally accessible via yaml.
kind: GrafanaAlertRule
kind: GrafanaAlertGroup
kind: GrafanaContactPoint
kind: GrafanaNotificationPolicy
Describe alternatives you've considered A workaround being considered is to manually mount the relevant config files into the running container via VolumeMount from a ConfigMap containing raw file contents. The only other alternative I am aware of is to manually enter these into the Grafana web console, which isn't the point.
Additional context Attached are screenshots of the relevant web console screens being referenced.
Existing solutions If applicable please provide a link to an existing solution from a different project
Work ongoing in https://github.com/grafana-operator/grafana-operator/pull/559
@NissesSenap @apryor6 how are those CRD's interacting with e.g. prometheus-operator/AlertmanagerConfig.
So after #559 got merged it's now possible to use unified alerting in the operator but we have no way to manage the config.
As pointed out by @apryor6 it would be nice to create a few new CRD:s to manage the new API. To move this issue forward we would need a CRD design. I have created a design example to manage organizations: https://github.com/grafana-operator/grafana-operator/pull/615, it would be good to have a similar design document for this issue. Is there any one that would be willing to start working on a design document? Just because you create the design it dosent mean today you have to implement the solution.
And as @Cellebyte points out should any on these new CRD:s be compatible with the Prometheus operator?
Would any one be willing to drive this forward?
I think that @apryor6 might have started to look in to a design proposal but I'm not sure. If you are willing to take a stab at a design document @R-Studio that would be great.
@NissesSenap Unfortunately, I don't think I have enough skills/experience to create a design document for this. (I've only been using Kubernetes/Container for 2 months😆)
Unified Alerting have been deprecated in grafana 9 and grafana alerting https://grafana.com/docs/grafana/latest/alerting/ is now the name of the game. https://grafana.com/blog/2021/06/14/the-new-unified-alerting-system-for-grafana-everything-you-need-to-know
This feature won't be implemented due to this. Since it's a allot of work and maintenance to take on a new controller. If you want grafana alerting CRD:s please create a new issue.
Apologies for not commenting recently. There's been no work done by me to create a design document. I'd really like to - it's purely been competing personal and professional priorities that make it hard.
IMO the approach would be to take the Grafana config spec for unified alerts and one-by-one capture those in the appropriate yaml spec. I understand from the Grafana team that by now there is full API support on the Grafana side for these, so surely it becomes viable to implement within the operator?