Updated rules using deprecated metrics with new versions.

[shubhambhattar]

Refer this changelog: https://github.com/kubernetes/kube-state-metrics/blob/33db2356bf1f0a1f51ddaaeb165bce04ab5aa0df/CHANGELOG.md#v200-alpha--2020-09-16

kube_pod_container_resource_requests_cpu_cores and kube_pod_container_resource_requests_memory_bytes are deprecated.

[brancz]

Any reason why we don't just include the memorySelector and byteSelector directly in this query? We try to keep configuration minimal if there aren't really any other ways to configure something.

[shubhambhattar]

We can. I thought this was the preferred way of doing it. I'll change that.

[brancz]

CI is failing, looks like our tests need some changes to the inputs since we're using different metrics with these changes. I think you should be fine just adapting the inputs of this test:

kubernetes-mixin/tests.yaml

Lines 113 to 156 in 3152600

	- interval: 1m
	input_series:
	- series: 'kube_pod_container_resource_requests_cpu_cores{container="kube-apiserver-67",endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",node="node-1",pod="pod-1",service="ksm"}'
	values: '0.15+0x10'
	- series: 'kube_pod_container_resource_requests_cpu_cores{container="kube-apiserver-67",endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",node="node-1",pod="pod-2",service="ksm"}'
	values: '0.15+0x10'
	- series: 'kube_pod_container_resource_requests_cpu_cores{container="kube-apiserver-67",endpoint="https-main",instance="ksm-2",job="kube-state-metrics",namespace="kube-apiserver",node="node-1",pod="pod-1",service="ksm"}'
	values: '0.1+0x10'
	- series: 'kube_pod_container_resource_requests_memory_bytes{container="kube-apiserver-67",endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",node="node-1",pod="pod-1",service="ksm"}'
	values: '1E9+0x10'
	- series: 'kube_pod_container_resource_requests_memory_bytes{container="kube-apiserver-67",endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",node="node-1",pod="pod-2",service="ksm"}'
	values: '1E9+0x10'
	- series: 'kube_pod_container_resource_requests_memory_bytes{container="kube-apiserver-67",endpoint="https-main",instance="ksm-2",job="kube-state-metrics",namespace="kube-apiserver",node="node-1",pod="pod-1",service="ksm"}'
	values: '0.5E9+0x10'
	# Duplicate kube_pod_status_phase timeseries for the same pod.
	- series: 'kube_pod_status_phase{endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",phase="Running",pod="pod-1",service="ksm"}'
	values: '1 stale'
	- series: 'kube_pod_status_phase{endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",phase="Pending",pod="pod-1",service="ksm"}'
	values: '1+0x10'
	- series: 'kube_pod_status_phase{endpoint="https-main",instance="ksm-1",job="kube-state-metrics",namespace="kube-apiserver",phase="Completed",pod="pod-2",service="ksm"}'
	values: '1+0x10'
	- series: 'kube_pod_status_phase{endpoint="https-main",instance="ksm-2",job="kube-state-metrics",namespace="kube-apiserver",phase="Running",pod="pod-1",service="ksm"}'
	values: '1+0x10'
	promql_expr_test:
	- eval_time: 0m
	expr: namespace:kube_pod_container_resource_requests_cpu_cores:sum
	exp_samples:
	- value: 0.15
	labels: 'namespace:kube_pod_container_resource_requests_cpu_cores:sum{namespace="kube-apiserver"}'
	- eval_time: 0m
	expr: namespace:kube_pod_container_resource_requests_memory_bytes:sum
	exp_samples:
	- value: 1.0e+9
	labels: 'namespace:kube_pod_container_resource_requests_memory_bytes:sum{namespace="kube-apiserver"}'
	- eval_time: 1m
	expr: namespace:kube_pod_container_resource_requests_cpu_cores:sum
	exp_samples:
	- value: 0.15
	labels: 'namespace:kube_pod_container_resource_requests_cpu_cores:sum{namespace="kube-apiserver"}'
	- eval_time: 1m
	expr: namespace:kube_pod_container_resource_requests_memory_bytes:sum
	exp_samples:
	- value: 1.0e+9
	labels: 'namespace:kube_pod_container_resource_requests_memory_bytes:sum{namespace="kube-apiserver"}'

[brancz]

Since these metrics have existed pretty much forever in kube-state-metrics, I'm happy with moving ahead right away with this.

Wdyt @paulfantom ?

[paulfantom]

We have similar PR in #534 and the result of discussion from it was to wait for a stable KSM release. Without such release, this PR is breaking compatibility for all users, and merging it will result in requiring mixed metrics coming from KSM v1 and v2.

I wouldn't have much objections if there would be any guarantee that stable KSM v2 would be released soon. However, as it is right now, the first alpha release of kube-state-metrics was on 2020-09-16 and this doesn't reassure me that stable release is going to be released in the next days.

[brancz]

I'm not sure I understand as the metrics as they are used in this PR have been around for years in kube-state-metric 1.x, only the ones currently used in the mixin are being dropped in 2.0. Am I missing something else?