EKS Karpenter

문학적인유사성·2024년 5월 26일
0

뎁옵깃옵쿠베

목록 보기
50/53
post-thumbnail

terraform에서는 그냥 helm module넣어서 바로 설치할수있지만,,, 여기는 terraform을 안쓰는 곳이여서... Helm으로 설치를해야한다..
기존에 아는것보다 더 알아가야 하니까.. 힘내자고
왜 예전에 스터디할때 기록을 안해두었을까...무조건 기록해야해...

CA를 쓸때는 autoscaling 그룹으로 요리조리 마음대로 하고싶을때 편한거 같고
Karpenter를 쓸때는, keda를 연결해서 keda가 pod를 조절하고 karpenter가 이리저리 따라갈때 좋은 거같다.
물론 속도는 karpenter가 훨씬 빠르다!
CA는 autoscaling그룹 조절하는 시간이 걸리기때문에, karpenter보다 느릴수밖에없지...!

karpenter를 쓸때는 infra nodegroup하나 만들어서, infra 관련 pod를 전부다 affinity나 nodeselector로 infra nodegroup에 띄워야한다. 그래야 나중에 대참사를 막을수있다..^^;;
생각하면 할수록... fargate로 띄우는게 진짜 좋은거같음...

0.36.2 올리면서 바뀐거 수정중


설치

export CLUSTER_NAME=goyo-eks
export AWS_DEFAULT_REGION=ap-northeast-2
AWS_ACCOUNT_ID=$(aws sts get-caller-identity --query Account --output text)
echo ${AWS_ACCOUNT_ID}

Karpenter Node Role 생성

아래의 권한을 넣어서 Role 생성

  • role-eks-node-goyo-karpenter
    신뢰정책
{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "sts:AssumeRole"
            ],
            "Principal": {
                "Service": [
                    "ec2.amazonaws.com"
                ]
            }
        }
    ]
}

policy

AmazonEKSWorkerNodePolicy
AmazonEKS_CNI_Policy
AmazonEC2ContainerRegistryReadOnly
AmazonSSMManagedInstanceCore

Karpenter Controller Role 생성

  • role-ekssa-goyo-karpenter
    policy 하나 만들어서 붙혀주기
{
    "Statement": [
        {
            "Action": [
                "ssm:GetParameter",
                "ec2:DescribeImages",
                "ec2:RunInstances",
                "ec2:DescribeSubnets",
                "ec2:DescribeSecurityGroups",
                "ec2:DescribeLaunchTemplates",
                "ec2:DescribeInstances",
                "ec2:DescribeInstanceTypes",
                "ec2:DescribeInstanceTypeOfferings",
                "ec2:DescribeAvailabilityZones",
                "ec2:DeleteLaunchTemplate",
                "ec2:CreateTags",
                "ec2:CreateLaunchTemplate",
                "ec2:CreateFleet",
                "ec2:DescribeSpotPriceHistory",
                "pricing:GetProducts"
            ],
            "Effect": "Allow",
            "Resource": "*",
            "Sid": "Karpenter"
        },
        {
            "Action": "ec2:TerminateInstances",
            "Condition": {
                "StringLike": {
                    "ec2:ResourceTag/karpenter.sh/nodepool": "*"
                }
            },
            "Effect": "Allow",
            "Resource": "*",
            "Sid": "ConditionalEC2Termination"
        },
        {
            "Effect": "Allow",
            "Action": "iam:PassRole",
            "Resource": "arn:aws:iam::${myaccountnumber}:role/role-eks-node-goyo-karpenter",
            "Sid": "PassNodeIAMRole"
        },
        {
            "Effect": "Allow",
            "Action": "eks:DescribeCluster",
            "Resource": "arn:aws:eks:ap-northeast-2:${myaccountnumber}:cluster/${CLUSTER_NAME}",
            "Sid": "EKSClusterEndpointLookup"
        },
        {
            "Sid": "AllowScopedInstanceProfileCreationActions",
            "Effect": "Allow",
            "Resource": "*",
            "Action": [
            "iam:CreateInstanceProfile"
            ],
            "Condition": {
            "StringEquals": {
                "aws:RequestTag/kubernetes.io/cluster/${CLUSTER_NAME}": "owned",
                "aws:RequestTag/topology.kubernetes.io/region": "ap-northeast-2"
            },
            "StringLike": {
                "aws:RequestTag/karpenter.k8s.aws/ec2nodeclass": "*"
            }
            }
        },
        {
            "Sid": "AllowScopedInstanceProfileTagActions",
            "Effect": "Allow",
            "Resource": "*",
            "Action": [
            "iam:TagInstanceProfile"
            ],
            "Condition": {
            "StringEquals": {
                "aws:ResourceTag/kubernetes.io/cluster/${CLUSTER_NAME}": "owned",
                "aws:ResourceTag/topology.kubernetes.io/region": "ap-northeast-2",
                "aws:RequestTag/kubernetes.io/cluster/${CLUSTER_NAME}": "owned",
                "aws:RequestTag/topology.kubernetes.io/region": "ap-northeast-2"
            },
            "StringLike": {
                "aws:ResourceTag/karpenter.k8s.aws/ec2nodeclass": "*",
                "aws:RequestTag/karpenter.k8s.aws/ec2nodeclass": "*"
            }
            }
        },
        {
            "Sid": "AllowScopedInstanceProfileActions",
            "Effect": "Allow",
            "Resource": "*",
            "Action": [
            "iam:AddRoleToInstanceProfile",
            "iam:RemoveRoleFromInstanceProfile",
            "iam:DeleteInstanceProfile"
            ],
            "Condition": {
            "StringEquals": {
                "aws:ResourceTag/kubernetes.io/cluster/${CLUSTER_NAME}": "owned",
                "aws:ResourceTag/topology.kubernetes.io/region": "ap-northeast-2"
            },
            "StringLike": {
                "aws:ResourceTag/karpenter.k8s.aws/ec2nodeclass": "*"
            }
            }
        },
        {
            "Sid": "AllowInstanceProfileReadActions",
            "Effect": "Allow",
            "Resource": "*",
            "Action": "iam:GetInstanceProfile"
        }
    ],
    "Version": "2012-10-17"
}
  • 신뢰관계
{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Federated": "arn:aws:iam::${AWS_ACCOUNT_ID}:oidc-provider/${oidc_endpoint}"
            },
            "Action": "sts:AssumeRoleWithWebIdentity",
            "Condition": {
                "StringEquals": {
                    "oidc.eks.ap-northeast-2.amazonaws.com/id/${oidc_endpoint}:aud": "sts.amazonaws.com",
                    "oidc.eks.ap-northeast-2.amazonaws.com/id/${oidc_endpoint}:sub": "system:serviceaccount:karpenter:karpenter"
                }
            }
        }
    ]
}

Subnet 태그 추가

$ aws eks list-nodegroups --cluster-name literary-goyo-eks --query 'nodegroups' --output text

$ aws eks describe-nodegroup --cluster-name literary-goyo-eks --nodegroup-name literary-goyo-eks-ng --query "nodegroup.subnets" --output text

해당 subnet에 태그 추가
key : karpenter.sh/discovery
value : literary-goyo-eks ( 넣고싶은 값 넣기, eks 클러스터 이름넣으면 편함 )

EKS SG 태그 추가

key : karpenter.sh/discovery
value : literary-goyo-eks (넣고 싶은 값 넣기, eks 클러스터 이름넣으면 편함 )

AWS auth 수정

kubectl edit configmap aws-auth -n kube-system

    - rolearn: aarn:aws:iam::${account_number}:role/${karpenternoderolename}
      username: system:node:{{EC2PrivateDNSName}}
      groups:
        - system:bootstrappers
        - system:nodes

Helm 설치

helm template karpenter oci://public.ecr.aws/karpenter/karpenter --version="0.36.2" --namespace "karpenter" \
    --set "settings.clusterName=literary-goyo-eks" \
    --set "serviceAccount.annotations.eks\.amazonaws\.com/role-arn"="arn:aws:iam::${myaccountnumber}:role/role-ekssa-goyo-karpenter" \
    --set "settings.isolatedVPC=true" \
    --set controller.resources.requests.cpu=1 \
    --set controller.resources.requests.memory=1Gi \
    --set controller.resources.limits.cpu=1 \
    --set controller.resources.limits.memory=1Gi  > karpenter.yaml
    
Deloyment.yaml에 affinity 추가 
카펜터가 만드는 노드에 카펜터가 상주하면, 큰일남..^^;;
나는 infranode에 다 띄울생각이여서 이름을 넣어줌 
            - matchExpressions:
              - key: eks.amazonaws.com/nodegroup
                operator: In
                values:
                - infra-nodegroup
                
kubectl create -f \
    https://raw.githubusercontent.com/aws/karpenter/${KARPENTER_VERSION}/pkg/apis/crds/karpenter.sh_provisioners.yaml
kubectl create -f \
    https://raw.githubusercontent.com/aws/karpenter/${KARPENTER_VERSION}/pkg/apis/crds/karpenter.k8s.aws_awsnodetemplates.yaml
kubectl create -f \
    https://raw.githubusercontent.com/aws/karpenter/${KARPENTER_VERSION}/pkg/apis/crds/karpenter.sh_machines.yaml
kubectl apply -f karpenter.yaml

kubectl api-resources \
    --categories karpenter \
    -o wide

0.36.2으로 오면서 provisionr, nodetemplate이 nodetemplate으로 바꼇네 ... ㅇㄴ 이거 바꿔야됨 ...^^; ; 그


---
apiVersion: karpenter.sh/v1beta1
kind: NodePool
metadata:
  name: default
spec:
  template:
    spec:
      requirements:
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64"]
        - key: kubernetes.io/os
          operator: In
          values: ["linux"]
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["on-demand"]
        - key: karpenter.k8s.aws/instance-category
          operator: In
          values: ["t"]
        - key: node.kubernetes.io/instance-type
          operator: In
          values: ["t3.medium"]
        - key: karpenter.k8s.aws/instance-generation
          operator: Gt
          values: ["2"]
      nodeClassRef:
        apiVersion: karpenter.k8s.aws/v1beta1
        kind: EC2NodeClass
        name: default
      # taints:
      # - key: karpenter
      #   value: "applied"
      #   effect: NoSchedule
  limits:
    cpu: 1000
  disruption:
    consolidationPolicy: WhenUnderutilized
    expireAfter: 720h # 30 * 24h = 720h
---
apiVersion: karpenter.k8s.aws/v1beta1
kind: EC2NodeClass
metadata:
  name: default
spec:
  amiFamily: AL2023
  role: role-eks-node-goyo-karpenter
  subnetSelectorTerms:
    - tags:
        karpenter.sh/discovery: literary-goyo-eks
  securityGroupSelectorTerms:
    - tags:
        karpenter.sh/discovery: literary-goyo-eks
  amiSelectorTerms:
    - id: ami-0f6b269f5b7aac774
  tags:
    project: goyo
    Env: test

테스트

apiVersion: apps/v1
kind: Deployment
metadata:
  creationTimestamp: null
  labels:
    app: nginx-test
  name: nginx-test
spec:
  replicas: 1
  selector:
    matchLabels:
      app: nginx-test
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: nginx-test
    spec:
      containers:
      - image: nginx
        name: nginx
        resources:
          requests:
            cpu: 20m
          limits:
            cpu: 100m
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: karpenter.sh/nodepool
                operator: In
                values:
                - default
kubectl logs karpenter-56fc488d76-5f96b -n karpenter

{"level":"INFO","time":"2024-05-31T05:15:05.839Z","logger":"controller.nodeclaim.lifecycle","message":"launched nodeclaim","commit":"e719109","nodeclaim":"default-7hw6r","provider-id":"aws:///ap-northeast-2a/i-00f498c84b028bead","instance-type":"t3.medium","zone":"ap-northeast-2a","capacity-type":"on-demand","allocatable":{"cpu":"1930m","ephemeral-storage":"17Gi","memory":"3246Mi","pods":"17"}}
{"level":"INFO","time":"2024-05-31T05:15:13.825Z","logger":"controller.provisioner","message":"found provisionable pod(s)","commit":"e719109","pods":"default/nginx-test-7f87977dc4-255w4","duration":"11.441124ms"}


kubectl get nodeclaim -w
NAME            TYPE        ZONE              NODE                                             READY   AGE
default-7hw6r   t3.medium   ap-northeast-2a   ip-10-10-3-178.ap-northeast-2.compute.internal   False   35s
default-7hw6r   t3.medium   ap-northeast-2a   ip-10-10-3-178.ap-northeast-2.compute.internal   False   40s
default-7hw6r   t3.medium   ap-northeast-2a   ip-10-10-3-178.ap-northeast-2.compute.internal   True    41s


kubectl get nodes
NAME                                              STATUS   ROLES    AGE   VERSION
ip-10-10-0-127.ap-northeast-2.compute.internal    Ready    <none>   41m   v1.29.3-eks-ae9a62a
ip-10-10-15-149.ap-northeast-2.compute.internal   Ready    <none>   41m   v1.29.3-eks-ae9a62a
ip-10-10-3-178.ap-northeast-2.compute.internal    Ready    <none>   54s   v1.29.0-eks-5e0fdde

30초안에 뜨는걸 볼수있음..!

pod를 삭제하면 karpenter에서 날려버린다.

kubectl logs karpenter-56fc488d76-5f96b -n karpenter
{"level":"INFO","time":"2024-05-31T05:17:23.379Z","logger":"controller.node.termination","message":"tainted node","commit":"e719109","node":"ip-10-10-3-178.ap-northeast-2.compute.internal"}
{"level":"INFO","time":"2024-05-31T05:17:23.397Z","logger":"controller.node.termination","message":"deleted node","commit":"e719109","node":"ip-10-10-3-178.ap-northeast-2.compute.internal"}
{"level":"INFO","time":"2024-05-31T05:17:23.814Z","logger":"controller.nodeclaim.termination","message":"deleted nodeclaim","commit":"e719109","nodeclaim":"default-7hw6r","node":"ip-10-10-3-178.ap-northeast-2.compute.internal","provider-id":"aws:///ap-northeast-2a/i-00f498c84b028bead"}
[

 kubectl get nodeclaim
No resources found

참고 terraform
# Allow terraform to access k8s, helm
provider "kubernetes" {
  host                   = module.eks.cluster_endpoint
  cluster_ca_certificate = base64decode(module.eks.cluster_certificate_authority_data)
  token                  = data.aws_eks_cluster_auth.eks.token
}
provider "helm" {
  kubernetes {
    host                   = module.eks.cluster_endpoint
    cluster_ca_certificate = base64decode(module.eks.cluster_certificate_authority_data)
    token                  = data.aws_eks_cluster_auth.eks.token
  }
}
data "aws_eks_cluster_auth" "eks" {name = module.eks.cluster_name}


resource "helm_release" "karpenter" {
  depends_on       = [module.eks.kubeconfig]
  namespace        = "karpenter"
  create_namespace = true

  name       = "karpenter"
  repository = "https://charts.karpenter.sh"
  chart      = "karpenter"
  version    = "v0.16.0"

  set {
    name  = "replicas"
    value = 2
  }

  set {
    name  = "serviceAccount.annotations.eks\\.amazonaws\\.com/role-arn"
    value = module.iam_assumable_role_karpenter.iam_role_arn
  }

  set {
    name  = "clusterName"
    value = local.cluster_name
  }

  set {
    name  = "clusterEndpoint"
    value = module.eks.cluster_endpoint
  }

  set {
    name  = "aws.defaultInstanceProfile"
    value = aws_iam_instance_profile.karpenter.name
  }
}
옛날 버전

### provisioner yaml
이거 중요함!
ttlSecondsAfterEmpty
ttlSecondsUntilExpired
consolidation

apiVersion: karpenter.k8s.aws/v1alpha1
kind: AWSNodeTemplate
metadata:
  name: application-nodetemplate
spec:
  subnetSelector:
    karpenter.sh/discovery: goyo-eks
  securityGroupSelector:
    karpenter.sh/discovery: goyo-eks
  instanceProfile: role-eks-node-goyo-karpenter
  amiFamily: AL2
  blockDeviceMappings:
    - deviceName: /dev/xvda
      ebs:
        volumeSize: 25G
        volumeType: gp3
        iops: 10000
        throughput: 125
        deleteOnTermination: true
  tags:
    Name : goyo-karpenter-test

### nodetemplate yaml

apiVersion: karpenter.k8s.aws/v1alpha1
kind: AWSNodeTemplate
metadata:
  name: application-nodetemplate
spec:
  subnetSelector:
    karpenter.sh/discovery: goyo-eks
  securityGroupSelector:
    karpenter.sh/discovery: goyo-eks
  instanceProfile: role-ekssa-goyo-karpenter
  amiFamily: AL2
  blockDeviceMappings:
    - deviceName: /dev/xvda
      ebs:
        volumeSize: 25G
        volumeType: gp3
        iops: 10000
        throughput: 125
        deleteOnTermination: true
  tags:
    Name : goyo-karpenter-test

참고

https://karpenter.sh/docs/getting-started/getting-started-with-karpenter/
https://devblog.kakaostyle.com/ko/2022-10-13-1-karpenter-on-eks/
https://aws-ia.github.io/terraform-aws-eks-blueprints/patterns/karpenter-mng/
https://techblog.gccompany.co.kr/karpenter-7170ae9fb677
https://younsl.github.io/blog/k8s/karpenter/

profile
Are you nervous? Don't be

0개의 댓글