Skip to content

Latest commit

 

History

History
97 lines (77 loc) · 11.5 KB

Structure.md

File metadata and controls

97 lines (77 loc) · 11.5 KB

Structure

The Kubernetes Monitoring Helm chart contains many software packages, and builds a comprehensive set of configuration and secrets for those packages. This document aims to describe and explain the structure of the Helm chart.

Kubernetes Monitoring inside of a Cluster

Software Deployed

This Helm chart deploys several packages to generate and capture the telemetry data on the cluster. This list corresponds to the list of dependencies in this chart's Chart.yaml file. For each package, there is an associated section inside the Helm chart's values.yaml file that controls how it is configured.

Name Type Associated values Description
Grafana Alloy StatefulSet alloy The Grafana Alloy instance that is responsible for scraping metrics, and accepting metrics, logs, and traces via receivers.
Grafana Alloy for Logs DaemonSet alloy-logs The Grafana Alloy instance that gathers Pod logs. By default, it uses HostPath volume mounts to read Pod log files directly from the nodes. It can alternatively get logs via the API server and be deployed as a Deployment.
Grafana Alloy for Events Deployment alloy-events The Grafana Alloy instance that is responsible for gathering Cluster events from the API server. This does not support clustering, so only one instance should be used.
Grafana Alloy for Profiles Deployment alloy-events The Grafana Alloy instance that is responsible for gathering profiles.
Grafana Alloy for Rules Deployment alloy-rules The Grafana Alloy instance that is responsible for synchronizing PrometheusRule objects to either Mimir or Loki.
kube-state-metrics Deployment kube-state-metrics A service for generating metrics about the state of the objects inside the Cluster.
Node Exporter DaemonSet prometheus-node-exporter An exporter used for gathering hardware and OS metrics for *NIX nodes of the Cluster.
Windows Exporter DaemonSet prometheus-windows-exporter An exporter used for gathering hardware and OS metrics for Windows nodes of the Cluster. Not deployed by default.
OpenCost Deployment opencost Used for gathering cost metrics for the Cluster.
Prometheus Operator CRDs CustomResourceDefinitions prometheus-operator-crds The custom resources for the Prometheus Operator. Use if you want to deploy PodMonitors, ServiceMonitors, or Probes.

Grafana Alloy instances

You may wonder why there are five instances of Grafana Alloy, rather than combining them. The reason is a balance between functionality and scalability. The default functionality of the Grafana Alloy for Logs is to gather logs via HostPath volume mounts. This requires it to be deployed as a DaemonSet. The Grafana Alloy for metrics and receivers is deployed as a StatefulSet, which allows it to be scaled (optionally with a HorizontalPodAutoscaler) based on load. If it was combined with the Alloy for Logs, it would lose its ability to scale.

Also, the Grafana Alloy for Events cannot be scaled beyond one replica, because that would result in duplicate events being sent to the logs service.

Configuration Created

This Helm chart also creates the configuration files, stored in ConfigMaps, for the Grafana Alloy instances. The following lists describe the pieces of the configuration files and the section in the chart's values.yaml file that controls it.

Grafana Alloy Configuration

Name Associated values Description
Annotation-based Autodiscovery .metrics.autoDiscover Controls how Pods and services are automatically discovered, based on their annotations.
Grafana Alloy .metrics.alloy Controls how to scrape metrics from the Grafana Alloy instances being deployed by this Helm chart.
kube-state-metrics .metrics.kube-state-metrics Controls how metrics from kube-state-metrics are discovered, scraped, and processed.
Node Exporter .metrics.node-exporter Controls how metrics from Node Exporter are discovered, scraped, and processed.
Windows Exporter .metrics.windows-exporter Controls how metrics from Windows Exporter are discovered, scraped, and processed.
Kubelet .metrics.kubelet Controls how metrics from the Kubelet are discovered, scraped, and processed.
cAdvisor .metrics.cadvisor Controls how metrics from cAdvisor are discovered, scraped, and processed.
API Server .metrics.apiserver Controls how metrics from the API Server are discovered, scraped, and processed.
Kube Controller Manager .metrics.kubeControllerManager Controls how metrics from the Kube Controller Manager are discovered, scraped, and processed.
Kube Proxy .metrics.kubeProxy Controls how metrics from the Kube Proxy are discovered, scraped, and processed.
Kube Scheduler .metrics.kubeScheduler Controls how metrics from the Kube Scheduler are discovered, scraped, and processed.
Cost Metrics .metrics.cost Controls how cost metrics are discovered, scraped, and processed.
PodMonitor Objects .metrics.podMonitors Controls how PodMonitor objects are discovered, scraped, and processed.
Probe Objects .metrics.Probes Controls how Probe objects are discovered, scraped, and processed.
ServiceMonitor Objects .metrics.serviceMonitors Controls how ServiceMonitor objects are discovered, scraped, and processed.
Kubernetes Monitoring .metrics.kubernetesMonitoring Enables or disables sending a static metric about how this Helm chart was deployed.
Filters for Received Metrics .metrics.receiver.filters Allows for filtering rules on metrics being received via the receivers.
Filters for Received Logs .logs.receiver.filters Allows for filtering rules on logs being received via the receivers.
Filters for Received Traces .traces.receiver.filters Allows for filtering rules on traces being received via the receivers.
Processors for Received data .receivers.processors Configuration for processors on telemetry data being received via the receivers.
Additional Configuration .extraConfig Additional configuration to be added to the Grafana Alloy.

Grafana Alloy for Events Configuration

Name Associated values Description
Cluster Events .logs.cluster_events Controls how cluster events are gathered.

Grafana Alloy for Logs Configuration

Name Associated values Description
Pod Logs .logs.pod_logs Controls how Pod logs are gathered.
PodLogs Objects .logs.podLogsObjects Controls how PodLogs objects are discovered and processed.
Additional Configuration .logs.extraConfig Additional configuration to be added to the Grafana Alloy instance for Logs.

Grafana Alloy for Profiles Configuration

Name Associated values Description
Profiles .profiles Controls how profiles are gathered.

Grafana Alloy for Rules

Name Associated values Description
Rules .rules Controls how PrometheusRule objects are discovered and synchronized.