Currently a lot of alerts relly on the job='kube-state-metrics' label.

Signed-off-by: Simon Pasquier <spasquie@redhat.com>

the addonresizer could lead to problems with the kube-state-metrics
it is also removed from the ksm maintained kubernetes manifests
https://github.com/kubernetes/kube-state-metrics/pull/750



Signed-off-by: Guus van Weelden <guus.vanweelden@moia.io>

These labels are confusing and misleading, as they describe
kube-state-metrics itself not the target itself.

Signed-off-by: yeya24 <yb532204897@gmail.com>

Signed-off-by: yeya24 <yb532204897@gmail.com>

release

Currently, we use the node:node_memory_bytes_total:sum and node:node_memory_bytes_available:sum
recording rules for the memory node query.
These recording rules have been removed in https://github.com/coreos/kube-prometheus/pull/191.

This fixes it by using raw queries.

`alertmanager-main` is hardcoded where it should derive from the real alertmanager name as defined by the user

If there is more than one prometheus-operator pod, which happens briefly
when we delete the prometheus-operator pod, we can see the errors of
many-to-many matching, this whitelists the labels matching, and excluded
the pod.

Signed-off-by: Guus van Weelden <guus.vanweelden@moia.io>

The resource is called storageclasses not storageclass.

This makes it easier to indetify which node alert is firing for.
Currently only device namespace and pod name were included in
the alert.

This allows us to display the node name in alerts, instead of the IP.

With the bump to kube-state-metrics v1.6 they added ingress and certificates but this updates the rbac rules so that those work with the new version.

Signed-off-by: Benjamin <benjamin@yunify.com>