2-installation.md

title
Installation

Install with Helm

If you have a Kubernetes cluster, Helm is the recommended installation method.

The following tutorial will guide you to install Karpor using Helm, which will install the chart with the release name karpor-release in namespace karpor.

Prerequisites

• Helm v3+
• A Kubernetes Cluster (The simplest way is to deploy a Kubernetes cluster locally using kind or minikube)

Remote Installation

First, add the karpor chart repo to your local repository.

helm repo add kusionstack https://kusionstack.github.io/charts
helm repo update

Then you can use the following command to install the latest version of Karpor.

helm install karpor-release kusionstack/karpor

Note that installing this chart directly means it will use the default template values for Karpor.

You may have to set your specific configurations if it is deployed into a production cluster, or you want to customize the chart configuration, such as resources, replicas, port etc.

All configurable parameters of the Karpor chart are detailed here.

helm install karpor-release kusionstack/karpor --set server.replicas=3 --set syncer.port=7654

Search all available versions

You can use the following command to view all installable chart versions.

helm repo update
helm search repo kusionstack/karpor --versions

Upgrade specified version

You can specify the version to be upgraded through the --version.

# Upgrade to the latest version.
helm upgrade karpor-release kusionstack/karpor

# Upgrade to the specified version.
helm upgrade karpor-release kusionstack/karpor --version 1.2.3

Local Installation

If you have problem connecting to https://kusionstack.github.io/charts/ in production, you may need to manually download the chart from here and use it to install or upgrade locally.

git clone https://github.com/KusionStack/charts.git
helm install karpor-release charts/karpor
helm upgrade karpor-release charts/karpor

Uninstall

To uninstall/delete the karpor-release Helm release in namespace karpor:

helm uninstall karpor-release

Image Registry Proxy for China

If you are in China and have problem to pull image from official DockerHub, you can use the registry proxy:

helm install karpor-release kusionstack/karpor --set registryProxy=docker.m.daocloud.io

NOTE: The above is just an example, you can replace the value of registryProxy as needed.

Enable AI features

If you want to install Karpor with AI features, including natural language search and AI analysis, you should configure parameters such as ai-auth-token, ai-base-url, etc., for example:

# Minimal configuration, using OpenAI as the default AI backend
helm install karpor-release kusionstack/karpor \
   --set server.ai.authToken={YOUR_AI_TOKEN}

# Example using Azure OpenAI
helm install karpor-release kusionstack/karpor \
   --set server.ai.authToken={YOUR_AI_TOKEN} \
   --set server.ai.baseUrl=https://{YOUR_RESOURCE_NAME}.openai.azure.com \
   --set server.ai.backend=azureopenai

# Example using Hugging Face
helm install karpor-release kusionstack/karpor \
   --set server.ai.authToken={YOUR_AI_TOKEN} \
   --set server.ai.model={YOUR_HUGGINGFACE_MODEL} \
   --set server.ai.backend=huggingface

# Custom configuration
helm install karpor-release kusionstack/karpor \
   --set server.ai.authToken={YOUR_AI_TOKEN} \
   --set server.ai.baseUrl=https://api.deepseek.com \
   --set server.ai.backend=openai \
   --set server.ai.model=deepseek-chat \
   --set server.ai.topP=0.5 \
   --set server.ai.temperature=0.2

# Example using AI Proxy
helm install karpor kusionstack/karpor \
   --set server.ai.authToken={YOUR_AI_TOKEN} \
   --set server.ai.proxy.enabled=true \
   --set server.ai.proxy.httpProxy={YOUR_HTTP_PROXY} \
   --set server.ai.proxy.httpsProxy={YOUR_HTTPS_PROXY} \
   --set server.ai.proxy.noProxy={YOUR_NO_PROXY}

Chart Parameters

The following table lists the configurable parameters of the chart and their default values.

General Parameters

Key	Type	Default	Description
namespace	string	"karpor"	Which namespace to be deployed.
namespaceEnabled	bool	true	Whether to generate namespace.
registryProxy	string	""	Image registry proxy will be the prefix as all component image.

Global Parameters

Key	Type	Default	Description
global.image.imagePullPolicy	string	"IfNotPresent"	Image pull policy to be applied to all Karpor components.

Karpor Server

The Karpor Server Component is main backend server. It itself is an apiserver, which also provides /rest-api to serve Dashboard.

Key	Type	Default	Description
server.ai	object	{"authToken":"","backend":"openai","baseUrl":"","model":"gpt-3.5-turbo","proxy":{"enabled":false,"httpProxy":"","httpsProxy":"","noProxy":""},"temperature":1,"topP":1}	AI configuration section. The AI analysis feature requires that [authToken, baseUrl] be assigned values.
server.ai.authToken	string	""	Authentication token for accessing the AI service.
server.ai.backend	string	"openai"	Backend service or platform that the AI model is hosted on. Available options: - "openai": OpenAI API (default) - "azureopenai": Azure OpenAI Service - "huggingface": Hugging Face API - "deepseek": DeepSeek API If the backend you are using is compatible with OpenAI, then there is no need to make any changes here.
server.ai.baseUrl	string	""	Base URL of the AI service. e.g., "https://api.openai.com/v1".
server.ai.model	string	"gpt-3.5-turbo"	Name or identifier of the AI model to be used. e.g., "gpt-3.5-turbo".
server.ai.proxy	object	{"enabled":false,"httpProxy":"","httpsProxy":"","noProxy":""}	Proxy configuration for AI service connections.
server.ai.proxy.enabled	bool	false	Enable proxy settings for AI service connections. When false, proxy settings will be ignored.
server.ai.proxy.httpProxy	string	""	HTTP proxy URL for AI service connections (e.g., "http://proxy.example.com:8080").
server.ai.proxy.httpsProxy	string	""	HTTPS proxy URL for AI service connections (e.g., "https://proxy.example.com:8080").
server.ai.proxy.noProxy	string	""	No proxy configuration for AI service connections (e.g., "localhost,127.0.0.1,example.com").
server.ai.temperature	float	1	Temperature parameter for the AI model. This controls the randomness of the output, where a higher value (e.g., 1.0) makes the output more random, and a lower value (e.g., 0.0) makes it more deterministic.
server.ai.topP	float	1	Top-p (nucleus sampling) parameter for the AI model. This controls Controls the probability mass to consider for sampling, where a higher value leads to greater diversity in the generated content (typically ranging from 0 to 1)
server.enableRbac	bool	false	Enable RBAC authorization if set to true.
server.image.repo	string	"kusionstack/karpor"	Repository for Karpor server image.
server.image.tag	string	""	Tag for Karpor server image. Defaults to the chart's appVersion if not specified.
server.name	string	"karpor-server"	Component name for karpor server.
server.port	int	7443	Port for karpor server.
server.replicas	int	1	The number of karpor server pods to run.
server.resources	object	{"limits":{"cpu":"500m","ephemeral-storage":"10Gi","memory":"1Gi"},"requests":{"cpu":"250m","ephemeral-storage":"2Gi","memory":"256Mi"}}	Resource limits and requests for the karpor server pods.
server.serviceType	string	"ClusterIP"	Service type for the karpor server. The available type values list as ["ClusterIP"、"NodePort"、"LoadBalancer"].

Karpor Syncer

The Karpor Syncer Component is independent server to synchronize cluster resources in real-time.

Key	Type	Default	Description
syncer.image.repo	string	"kusionstack/karpor"	Repository for Karpor syncer image.
syncer.image.tag	string	""	Tag for Karpor syncer image. Defaults to the chart's appVersion if not specified.
syncer.name	string	"karpor-syncer"	Component name for Karpor syncer.
syncer.port	int	7443	Port for Karpor syncer.
syncer.replicas	int	1	The number of karpor syncer pods to run.
syncer.resources	object	{"limits":{"cpu":"500m","ephemeral-storage":"10Gi","memory":"1Gi"},"requests":{"cpu":"250m","ephemeral-storage":"2Gi","memory":"256Mi"}}	Resource limits and requests for the karpor syncer pods.

ElasticSearch

The ElasticSearch Component to store the synchronized resources and user data.

Key	Type	Default	Description
elasticsearch.image.repo	string	"docker.elastic.co/elasticsearch/elasticsearch"	Repository for ElasticSearch image.
elasticsearch.image.tag	string	"8.6.2"	Specific tag for ElasticSearch image.
elasticsearch.name	string	"elasticsearch"	Component name for ElasticSearch.
elasticsearch.port	int	9200	Port for ElasticSearch.
elasticsearch.replicas	int	1	The number of ElasticSearch pods to run.
elasticsearch.resources	object	{"limits":{"cpu":"2","ephemeral-storage":"10Gi","memory":"4Gi"},"requests":{"cpu":"2","ephemeral-storage":"10Gi","memory":"4Gi"}}	Resource limits and requests for the karpor elasticsearch pods.

ETCD

The ETCD Component is the storage of Karpor Server as apiserver.

Key	Type	Default	Description
etcd.image.repo	string	"quay.io/coreos/etcd"	Repository for ETCD image.
etcd.image.tag	string	"v3.5.11"	Specific tag for ETCD image.
etcd.name	string	"etcd"	Component name for ETCD.
etcd.persistence.accessModes	list	["ReadWriteOnce"]	Volume access mode, ReadWriteOnce means single node read-write access
etcd.persistence.size	string	"10Gi"	Size of etcd persistent volume
etcd.port	int	2379	Port for ETCD.
etcd.replicas	int	1	The number of etcd pods to run.
etcd.resources	object	{"limits":{"cpu":"500m","ephemeral-storage":"10Gi","memory":"1Gi"},"requests":{"cpu":"250m","ephemeral-storage":"2Gi","memory":"256Mi"}}	Resource limits and requests for the karpor etcd pods.

Job

This one-time job is used to generate root certificates and some preliminary work.

Key	Type	Default	Description
job.image.repo	string	"kusionstack/karpor"	Repository for the Job image.
job.image.tag	string	""	Tag for Karpor image. Defaults to the chart's appVersion if not specified.