Comprender el servicio kube-state-metrics
Nota:
kube-state-metrics es compatible como una característica solo beta.
kube-state-metrics es un servicio de monitoreo de métricas de terceros que genera métricas basadas en el estado actual de los clústeres de Kubernetes. Puede usar kube-state-metrics para monitorear el estado del clúster y los servicios de Kubernetes. El servicio kube-state-metrics es compatible como una característica solo beta. kube-state-metrics se ejecuta como un servicio de clúster y se instala automáticamente al instalar Paragon Automation. Una vez instalado este servicio, puede habilitar este servicio para generar, monitorear y exponer métricas de varios objetos dentro de un clúster de Kubernetes.
El servicio kube-state-metrics proporciona métricas sobre pods, DaemonSets, despliegues, volumen persistente, puntos finales, entrada, trabajo, arrendamiento y objetos configmap que forman parte de un clúster de Kubernetes.
Lista de métricas expuestas
La siguiente es la lista de métricas que se exponen:
- Pods que se ejecutan en un espacio de nombres
- Pods disponibles
- Información sobre implementaciones exitosas o fallidas
- Estado de los volúmenes persistentes
- Información sobre trabajos en ejecución, exitosos y fallidos
- Pods que están en estado de error
- Salud de la implementación y DaemonSets
- Estado y condición de los nodos de Kubernetes
Habilitar kube-state-metrics
Puede realizar la habilitación kube-state-metrics desde la CLI y desde la interfaz de usuario de Paragon Automation.
La siguiente es una descripción general de los pasos para habilitar kube-state-metrics desde la interfaz de usuario:
-
Cree un dispositivo no administrado.
El dispositivo no administrado representa el clúster. El nombre de host del dispositivo debe ser la IP del servicio kube-state-metrics. Por ejemplo,
kube-state-metrics.healthbot.svc.cluster.local. -
Agregar el dispositivo a un grupo de dispositivos.
-
Crear reglas.
-
Aplicar playbooks.
Ejemplo de reglas y manuales
check-daemonset-status.rule
healthbot {
topic kube-metrics {
rule check-daemonset-status {
keys [ daemonset namespace ];
synopsis "";
description "Checks daemon set unavailable status";
sensor daemonset-status {
description "Checks daemon set unavailable status";
server-monitoring {
sensor-name /kube/daemonset;
frequency 60s;
}
}
field daemonset {
sensor daemonset-status {
path daemonset;
}
type string;
description "Checks status for demonset key";
}
field daemonset_status {
sensor daemonset-status {
path /kube/daemonset/status/number/unavailable;
}
type float;
description "Field to check condition";
}
field namespace {
sensor daemonset-status {
path namespace;
}
type string;
description "Checks status for namespace key";
}
trigger daemonset-status {
frequency 1offset;
term available {
when {
equal-to "$daemonset_status" 0;
}
then {
status {
color green;
message "Unavailable demons set is 0 for $namespace $daemonset";
}
}
}
term notavailable {
then {
status {
color red;
message "Unavailable demons set is not 0 for $namespace $daemonset";
}
}
}
}
}
}
}
check-deployment-status-condition.rule
healthbot {
topic kube-metrics {
rule check-deployment-status-condition {
keys [ condition deployment namespace ];
synopsis "";
description "Checks kube metrics deployment status condition";
sensor deployment-status {
description "Checks kube metrics deployment status condition";
server-monitoring {
sensor-name /kube/deployment;
frequency 60s;
}
}
field condition {
sensor deployment-status {
path condition;
}
type string;
description "Deployment condition";
}
field deployment {
sensor deployment-status {
path deployment;
}
type string;
description "Deployment pod name";
}
field namespace {
sensor deployment-status {
path namespace;
}
type string;
description "Deployment namespace";
}
field status {
sensor deployment-status {
path status;
}
type string;
description "Checks for true or false condition";
}
trigger deployment-status {
frequency 1offset;
term available {
when {
matches-with "$status" true;
}
then {
status {
color green;
message "Deployment status for $deployment is $status";
}
}
}
term notavailable {
then {
status {
color red;
message "Deployment status for $deployment is $status";
}
}
}
}
}
}
}
check-deployment-status-replicas.rule
healthbot {
topic kube-metrics {
rule check-deployment-status-replicas {
keys [ deployment namespace ];
synopsis "";
description "Checks kube metrics deployment replica status";
sensor deployment-status {
description "Checks kube metrics deployment replica status";
server-monitoring {
sensor-name /kube/deployment;
frequency 60s;
}
}
field deployment {
sensor deployment-status {
path deployment;
}
type string;
description "Deployment pod name";
}
field deployment_status {
sensor deployment-status {
path /kube/deployment/status/replicas;
}
type float;
description "Field to check 0 or other values";
}
field namespace {
sensor deployment-status {
path namespace;
}
type string;
description "Namespace key";
}
trigger deployment-status {
frequency 1offset;
term available {
when {
not-equal-to "$deployment_status" 0;
}
then {
status {
color green;
message "Deployment status for replicaset is $deployment_status for $namespace $deployment";
}
}
}
term notavailable {
then {
status {
color red;
message "Deployment status for replicaset is $deployment_status for $namespace $deployment";
}
}
}
}
}
}
}
check-node-status.rule
healthbot {
topic kube-metrics {
rule check-node-status {
keys [ condition node ];
synopsis "";
description "Checks node status";
sensor node-status {
description "Checks node status";
server-monitoring {
sensor-name /kube/node;
frequency 60s;
}
}
field condition {
sensor node-status {
path condition;
}
type string;
description "Node condition";
}
field node {
sensor node-status {
path node;
}
type string;
description "Node name";
}
field node_status {
constant {
value "{{value}}";
}
type integer;
description "Field to check condition";
}
field status {
sensor node-status {
path status;
}
type string;
description "Status of the node";
}
trigger node-status {
frequency 1offset;
term available {
when {
matches-with "$status" false {
ignore-case;
}
}
then {
status {
color green;
message "$condition for $node is $status";
}
}
}
term notavailable {
then {
status {
color red;
message "$condition for $node is $status.";
}
}
}
}
variable value {
value 0;
description "Variable to match true(0) condition.";
type int;
}
}
}
}
check-pod-container-restarts.rule
healthbot {
topic kube-metrics {
rule check-pod-container-restarts {
keys [ container namespace pod uid ];
synopsis "";
description "Checks pod container status restarts";
sensor pod-init-container-status {
description "Checks pod container status restarts";
server-monitoring {
sensor-name /kube/pod/container;
frequency 60s;
}
}
field container {
sensor pod-init-container-status {
path container;
}
type string;
description "container name";
}
field namespace {
sensor pod-init-container-status {
path namespace;
}
type string;
description "namespace of the pod";
}
field pod {
sensor pod-init-container-status {
path pod;
}
type string;
description "pod name";
}
field restart-value {
sensor pod-init-container-status {
path /kube/pod/container/status/restarts/total;
}
type integer;
description "restart status value";
}
field uid {
sensor pod-init-container-status {
path uid;
}
type string;
description "Id ";
}
trigger restart-total-status {
frequency 1offset;
term less-than-ten {
when {
less-than-or-equal-to "$restart-value" 10;
}
then {
status {
color green;
message "$namespace pod $pod container $container restart total is $restart-value ";
}
}
}
term between-ten-and-twenty {
when {
range "$restart-value" {
min 10;
max 20;
}
}
then {
status {
color yellow;
message "$namespace pod $pod container $container restart total is $restart-value ";
}
}
}
term more-than-twenty {
then {
status {
color red;
message "$namespace pod $pod container $container restart total is $restart-value";
}
}
}
}
}
}
}
check-pod-init-container-status.rule
healthbot {
topic kube-metrics {
rule check-pod-init-container-status {
keys [ container namespace pod uid ];
synopsis "";
description "Checks pod init container status waiting";
sensor pod-init-container-status {
description "Checks pod init container status waiting";
server-monitoring {
sensor-name /kube/pod/init/container;
frequency 60s;
}
}
field container {
sensor pod-init-container-status {
path container;
}
type string;
description "container name";
}
field namespace {
sensor pod-init-container-status {
path namespace;
}
type string;
description "namespace of the pod";
}
field pod {
sensor pod-init-container-status {
path pod;
}
type string;
description "pod name";
}
field status {
sensor pod-init-container-status {
path /kube/pod/init/container/status/waiting;
}
type integer;
description "Statius value (0/1)";
}
field uid {
sensor pod-init-container-status {
path uid;
}
type string;
description "Id ";
}
trigger waiting-status {
frequency 1offset;
term matches-zero {
when {
equal-to "$status" 0 {
time-range 10offset;
}
}
then {
status {
color green;
message "$namespace $pod $container status is $status";
}
}
}
term is-not-zero {
then {
status {
color red;
message "$namespace $pod $container status is $status";
}
}
}
}
}
}
}
kube-metrics.playbook
healthbot {
playbook kube-metrics {
rules [ kube-metrics/check-daemonset-status kube-metrics/check-deployment-status-replicas kube-metrics/check-node-status kube-metrics/check-pod-container-restarts kube-metrics/check-pod-init-container-status kube-metrics/check-deployment-status-condition ];
description "Rules to check for kube-metrics ";
synopsis "Rules to check for kube-metrics ";
}
}