Add "Terminating" status in kube_pod_status_phase metrics

Makefile

@@ -49,7 +49,7 @@ doccheck: generate
 	@echo "- Checking if the generated documentation is up to date..."
 	@git diff --exit-code
 	@echo "- Checking if the documentation is in sync with the code..."
-	@grep -hoE '(kube_[^ |]+)' docs/* --exclude=README.md| sort -u > documented_metrics
+	@grep -hoE '(\| kube_[^ |]+)' docs/* --exclude=README.md| sed -E 's/\| //g' | sort -u > documented_metrics
 	@find internal/store -type f -not -name '*_test.go' -exec sed -nE 's/.*"(kube_[^"]+)"/\1/p' {} \; | sed -E 's/,//g' | sort -u > code_metrics
 	@diff -u0 code_metrics documented_metrics || (echo "ERROR: Metrics with - are present in code but missing in documentation, metrics with + are documented but not found in code."; exit 1)
 	@echo OK

docs/pod-metrics.md

@@ -21,7 +21,8 @@
 | kube_pod_container_status_restarts_total | Counter | `container`=&lt;container-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `pod`=&lt;pod-name&gt; | STABLE |
 | kube_pod_container_resource_requests | Gauge | `resource`=&lt;resource-name&gt; <br> `unit`=&lt;resource-unit&gt; <br> `container`=&lt;container-name&gt; <br> `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `node`=&lt; node-name&gt; | STABLE |
 | kube_pod_container_resource_limits | Gauge | `resource`=&lt;resource-name&gt; <br> `unit`=&lt;resource-unit&gt; <br> `container`=&lt;container-name&gt; <br> `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `node`=&lt; node-name&gt; | STABLE |
-| kube_pod_created | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; |
+| kube_pod_created | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; | STABLE |
+| kube_pod_deleted | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; | EXPERIMENTAL |
 | kube_pod_restart_policy | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `type`=&lt;Always|Never|OnFailure&gt; | STABLE |
 | kube_pod_init_container_info | Gauge | `container`=&lt;container-name&gt; <br> `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `image`=&lt;image-name&gt; <br> `image_id`=&lt;image-id&gt; <br> `container_id`=&lt;containerid&gt; | STABLE |
 | kube_pod_init_container_status_waiting | Gauge | `container`=&lt;container-name&gt; <br> `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; | STABLE |
@@ -35,5 +36,35 @@
 | kube_pod_init_container_resource_limits | Gauge | `resource`=&lt;resource-name&gt; <br> `unit`=&lt;resource-unit&gt; <br> `container`=&lt;container-name&gt; <br> `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `node`=&lt; node-name&gt; | STABLE |
 | kube_pod_spec_volumes_persistentvolumeclaims_info | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `volume`=&lt;volume-name&gt;  <br> `persistentvolumeclaim`=&lt;persistentvolumeclaim-claimname&gt; | STABLE |
 | kube_pod_spec_volumes_persistentvolumeclaims_readonly | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt;  <br> `volume`=&lt;volume-name&gt;  <br> `persistentvolumeclaim`=&lt;persistentvolumeclaim-claimname&gt; | STABLE |
+| kube_pod_status_reason | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; <br> `reason`=&lt;NodeLost\|Evicted\&gt; | EXPERIMENTAL |
 | kube_pod_status_scheduled_time | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; | STABLE |
 | kube_pod_status_unschedulable | Gauge | `pod`=&lt;pod-name&gt; <br> `namespace`=&lt;pod-namespace&gt; | STABLE |
+
+## Useful metrics queries
+
+### How to retrieve none standard Pod state
+
+It is not straightforward to get the Pod states for certain cases like "Terminating" and "Unknow" since it is not stored behind a field in the `Pod.status`.
+
+So to get them, you need to compose multiple metrics (like it is done in the `kubectl` command line).
+
+For example:
+
+* To get the list the Pods that are in an `Unknow` state you can to the following promQL query: `count(kube_pod_status_phase{phase="Running"}) by (namespace, pod) * count(kube_pod_status_reason{reason="NodeLost"}) by(namespace, pod)`
+
+* For Pods in `Terminated` state: `count(kube_pod_status_phase{phase="Running"}) by (namespace, pod) * count(kube_pod_deleted) by (namespace, pod) * count(kube_pod_status_reason{reason!="NodeLost"})) by (namespace, pod)`
+
+A useful "Rule" can be to be alerted when a Pod is blocked in `Terminated` state for more than `5m`.
+
+```yaml
+groups:
+- name: Pod state
+  rules:
+  - alert: PodsBlockInTerminatingState
+    expr: count(kube_pod_status_phase{phase="Running"}) by (namespace, pod) * count(kube_pod_deleted) by (namespace, pod) * count(kube_pod_status_reason{reason!="NodeLost"})) by (namespace, pod) > 0
+    for: 5m
+    labels:
+      severity: page
+    annotations:
+      summary: Pod {{labels.namespace}}/{{labels.pod}} block in terminating state.
+```

internal/store/pod.go

@@ -37,6 +37,7 @@ var (
 	descPodLabelsDefaultLabels = []string{"namespace", "pod"}
 	containerWaitingReasons    = []string{"ContainerCreating", "CrashLoopBackOff", "CreateContainerConfigError", "ErrImagePull", "ImagePullBackOff", "CreateContainerError", "InvalidImageName"}
 	containerTerminatedReasons = []string{"OOMKilled", "Completed", "Error", "ContainerCannotRun", "DeadlineExceeded", "Evicted"}
+	podStatusReasons           = []string{"NodeLost", "Evicted"}
 
 	podMetricFamilies = []generator.FamilyGenerator{
 		{
@@ -197,6 +198,26 @@ var (
 				}
 			}),
 		},
+		{
+			Name: "kube_pod_deleted",
+			Type: metric.Gauge,
+			Help: "Unix deletion timestamp",
+			GenerateFunc: wrapPodFunc(func(p *v1.Pod) *metric.Family {
+				ms := []*metric.Metric{}
+
+				if p.DeletionTimestamp != nil && !p.DeletionTimestamp.IsZero() {
+					ms = append(ms, &metric.Metric{
+						LabelKeys:   []string{},
+						LabelValues: []string{},
+						Value:       float64(p.DeletionTimestamp.Unix()),
+					})
+				}
+
+				return &metric.Family{
+					Metrics: ms,
+				}
+			}),
+		},
 		{
 			Name: "kube_pod_restart_policy",
 			Type: metric.Gauge,
@@ -354,6 +375,30 @@ var (
 				}
 			}),
 		},
+		{
+			Name: "kube_pod_status_reason",
+			Type: metric.Gauge,
+			Help: "The pod status reasons",
+			GenerateFunc: wrapPodFunc(func(p *v1.Pod) *metric.Family {
+				ms := []*metric.Metric{}
+
+				for _, reason := range podStatusReasons {
+					metric := &metric.Metric{}
+					metric.LabelKeys = []string{"reason"}
+					metric.LabelValues = []string{reason}
+					if p.Status.Reason == reason {
+						metric.Value = boolFloat64(true)
+					} else {
+						metric.Value = boolFloat64(false)
+					}
+					ms = append(ms, metric)
+				}
+
+				return &metric.Family{
+					Metrics: ms,
+				}
+			}),
+		},
 		{
 			Name: "kube_pod_container_info",
 			Type: metric.Gauge,

internal/store/pod_test.go

@@ -863,6 +863,32 @@ kube_pod_container_status_last_terminated_reason{container="container7",namespac
 `,
 			MetricNames: []string{"kube_pod_created", "kube_pod_info", "kube_pod_start_time", "kube_pod_completion_time", "kube_pod_owner"},
 		},
+		{
+			Obj: &v1.Pod{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:              "pod1",
+					CreationTimestamp: metav1.Time{Time: time.Unix(1500000000, 0)},
+					Namespace:         "ns1",
+					UID:               "abc-123-xxx",
+					DeletionTimestamp: &metav1.Time{Time: time.Unix(1800000000, 0)},
+				},
+				Spec: v1.PodSpec{
+					NodeName:          "node1",
+					PriorityClassName: "system-node-critical",
+				},
+				Status: v1.PodStatus{
+					HostIP:    "1.1.1.1",
+					PodIP:     "1.2.3.4",
+					StartTime: &metav1StartTime,
+				},
+			},
+			Want: `
+				# HELP kube_pod_deleted Unix deletion timestamp
+				# TYPE kube_pod_deleted gauge
+				kube_pod_deleted{namespace="ns1",pod="pod1"} 1.8e+09
+`,
+			MetricNames: []string{"kube_pod_deleted"},
+		},
 		{
 			Obj: &v1.Pod{
 				ObjectMeta: metav1.ObjectMeta{
@@ -1055,14 +1081,58 @@ kube_pod_container_status_last_terminated_reason{container="container7",namespac
 			},
 			Want: `
 				# HELP kube_pod_status_phase The pods current phase.
+				# HELP kube_pod_status_reason The pod status reasons
 				# TYPE kube_pod_status_phase gauge
+				# TYPE kube_pod_status_reason gauge
 				kube_pod_status_phase{namespace="ns4",phase="Failed",pod="pod4"} 0
 				kube_pod_status_phase{namespace="ns4",phase="Pending",pod="pod4"} 0
 				kube_pod_status_phase{namespace="ns4",phase="Running",pod="pod4"} 0
 				kube_pod_status_phase{namespace="ns4",phase="Succeeded",pod="pod4"} 0
 				kube_pod_status_phase{namespace="ns4",phase="Unknown",pod="pod4"} 1
+				kube_pod_status_reason{namespace="ns4",pod="pod4",reason="Evicted"} 0
+				kube_pod_status_reason{namespace="ns4",pod="pod4",reason="NodeLost"} 1
 `,
-			MetricNames: []string{"kube_pod_status_phase"},
+			MetricNames: []string{"kube_pod_status_phase", "kube_pod_status_reason"},
+		},
+		{
+			Obj: &v1.Pod{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:              "pod4",
+					Namespace:         "ns4",
+					DeletionTimestamp: &metav1.Time{},
+				},
+				Status: v1.PodStatus{
+					Phase:  v1.PodRunning,
+					Reason: "Evicted",
+				},
+			},
+			Want: `
+				# HELP kube_pod_status_reason The pod status reasons
+				# TYPE kube_pod_status_reason gauge
+				kube_pod_status_reason{namespace="ns4",pod="pod4",reason="Evicted"} 1
+				kube_pod_status_reason{namespace="ns4",pod="pod4",reason="NodeLost"} 0
+`,
+			MetricNames: []string{"kube_pod_status_reason"},
+		},
+		{
+			Obj: &v1.Pod{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:              "pod4",
+					Namespace:         "ns4",
+					DeletionTimestamp: &metav1.Time{},
+				},
+				Status: v1.PodStatus{
+					Phase:  v1.PodRunning,
+					Reason: "other reason",
+				},
+			},
+			Want: `
+				# HELP kube_pod_status_reason The pod status reasons
+				# TYPE kube_pod_status_reason gauge
+				kube_pod_status_reason{namespace="ns4",pod="pod4",reason="Evicted"} 0
+				kube_pod_status_reason{namespace="ns4",pod="pod4",reason="NodeLost"} 0
+`,
+			MetricNames: []string{"kube_pod_status_reason"},
 		},
 		{
 			Obj: &v1.Pod{
@@ -1535,7 +1605,7 @@ func BenchmarkPodStore(b *testing.B) {
 		},
 	}
 
-	expectedFamilies := 35
+	expectedFamilies := 41
 	for n := 0; n < b.N; n++ {
 		families := f(pod)
 		if len(families) != expectedFamilies {

main_test.go

@@ -171,6 +171,8 @@ kube_pod_labels{namespace="default",pod="pod0"} 1
 # HELP kube_pod_created Unix creation timestamp
 # TYPE kube_pod_created gauge
 kube_pod_created{namespace="default",pod="pod0"} 1.5e+09
+# HELP kube_pod_deleted Unix deletion timestamp
+# TYPE kube_pod_deleted gauge
 # HELP kube_pod_restart_policy Describes the restart policy in use by this pod.
 # TYPE kube_pod_restart_policy gauge
 kube_pod_restart_policy{namespace="default",pod="pod0",type="Always"} 1
@@ -187,6 +189,10 @@ kube_pod_status_phase{namespace="default",pod="pod0",phase="Running"} 1
 kube_pod_status_phase{namespace="default",pod="pod0",phase="Unknown"} 0
 # HELP kube_pod_status_ready Describes whether the pod is ready to serve requests.
 # TYPE kube_pod_status_ready gauge
+# HELP kube_pod_status_reason The pod status reasons
+# TYPE kube_pod_status_reason gauge
+kube_pod_status_reason{namespace="default",pod="pod0",reason="Evicted"} 0
+kube_pod_status_reason{namespace="default",pod="pod0",reason="NodeLost"} 0
 # HELP kube_pod_status_scheduled Describes the status of the scheduling process for the pod.
 # TYPE kube_pod_status_scheduled gauge
 # HELP kube_pod_container_info Information about a container in a pod.