Add kubeflow applicationset

This commit is contained in:
wbsong111
2025-06-25 14:09:06 +09:00
parent c2dc36ea75
commit 24c017faab
9 changed files with 250 additions and 0 deletions
+75
View File
@@ -0,0 +1,75 @@
# ArgoCD를 이용한 Kubeflow 배포
이 저장소에는 **ApplicationSet을 사용한 Kubeflow 배포** 구현이 포함되어 있습니다.
### 배포 방법
1. ApplicationSet 적용:
```bash
kubectl apply -f argocd-solutions/applicationset/kubeflow-applicationset.yaml
```
2. ArgoCD UI에서 배포 상태 확인:
- `istio-system` → `kubeflow-dependencies` → `kubeflow-core` 순서로 배포
- 각 단계별 동기화 상태 모니터링
### 배포 구조
ApplicationSet은 다음과 같은 3단계 배포 순서로 구성되어 있습니다:
#### Wave 0: Istio System (`istio-system`)
- **네임스페이스**: `istio-system`
- **의존성**: 없음 (최우선 배포)
- **구성 요소**:
- Istio CRDs
- Istio Namespace
- Istio 설치 (OAuth2-Proxy 오버레이)
- 클러스터 로컬 게이트웨이
#### Wave 1: Kubeflow Dependencies (`kubeflow-dependencies`)
- **네임스페이스**: `kubeflow`
- **의존성**: `istio-system`
- **구성 요소**:
- Cert-Manager (Kubeflow Issuer)
- OAuth2-Proxy (M2M Keycloak 오버레이)
- KNative Serving (Gateway 오버레이)
- Kubeflow Namespace
- Network Policies
- Kubeflow Roles
- Kubeflow Istio Resources
#### Wave 2: Kubeflow Core (`kubeflow-core`)
- **네임스페이스**: `kubeflow`
- **의존성**: `kubeflow-dependencies`
- **구성 요소**:
- Kubeflow Pipelines (Multi-user, Cert-Manager)
- Katib (Kubeflow 통합)
- Central Dashboard (OAuth2-Proxy)
- Admission Webhook (Cert-Manager)
- Jupyter Web App & Notebook Controller
- Profiles + KFAM
- PVC Viewer & Volumes Web App
- Tensorboard Controller & Web App
- Training Operator
- KServe & Models Web App
- Spark Operator
- User Namespace
- Kubeflow Ingress
### 주요 특징
1. **순차적 배포**: Sync Wave를 통한 명확한 배포 순서 보장
2. **자동화된 동기화**:
- 자동 Prune 및 Self-Heal 활성화
- Server-Side Apply 사용
- 네임스페이스 자동 생성
3. **재시도 정책**: 실패 시 지수 백오프로 최대 5회 재시도
4. **의존성 추적**: 각 Application의 의존성 정보를 메타데이터로 관리
### 장점
- **중앙집중식 관리**: 하나의 ApplicationSet으로 전체 Kubeflow 스택 관리
- **명확한 의존성**: Wave 기반 순차 배포로 의존성 문제 해결
- **확장성**: 새로운 컴포넌트 추가 시 리스트에 항목만 추가
- **일관성**: 모든 Application이 동일한 정책과 설정 적용
@@ -0,0 +1,59 @@
# ApplicationSet을 사용한 방법
apiVersion: argoproj.io/v1alpha1
kind: ApplicationSet
metadata:
name: kubeflow-applicationset
namespace: platform
spec:
generators:
- list:
elements:
- name: istio-system
path: kubeflow/dip/istio-system
namespace: istio-system
wave: "0"
dependencies: ""
- name: kubeflow-dependencies
path: kubeflow/dip/kubeflow-dependencies
namespace: kubeflow
wave: "1"
dependencies: "istio-system"
- name: kubeflow-core
path: kubeflow/dip/kubeflow-core
namespace: kubeflow
wave: "2"
dependencies: "kubeflow-dependencies"
template:
metadata:
name: '{{name}}'
annotations:
argocd.argoproj.io/sync-wave: '{{wave}}'
finalizers:
- resources-finalizer.argocd.argoproj.io
spec:
project: default
source:
repoURL: https://gitea.example.org/dip/tenant-catalog
targetRevision: kubeflow/v1.10.0
path: '{{path}}'
destination:
server: https://kubernetes.default.svc
namespace: '{{namespace}}'
syncPolicy:
automated:
prune: true
selfHeal: true
syncOptions:
- CreateNamespace=true
- ServerSideApply=true
retry:
limit: 5
backoff:
duration: 5s
factor: 2
maxDuration: 3m
info:
- name: 'Wave'
value: '{{wave}}'
- name: 'Dependencies'
value: '{{dependencies}}'
@@ -0,0 +1,31 @@
# Istio System Kustomization
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
# 배포 순서 정의
sortOptions:
order: legacy
legacySortOptions:
orderFirst:
- Namespace
- CustomResourceDefinition
- ServiceAccount
- ClusterRole
- ClusterRoleBinding
- ConfigMap
- Secret
- Service
orderLast:
- ValidatingWebhookConfiguration
- MutatingWebhookConfiguration
- Deployment
resources:
# 1. CRDs 먼저
- ../../common/istio-1-24/istio-crds/base
# 2. Namespace
- ../../common/istio-1-24/istio-namespace/base
# 3. Istio 설치
- ../../common/istio-1-24/istio-install/overlays/oauth2-proxy
# 4. 클러스터 로컬 게이트웨이
- ../../common/istio-1-24/cluster-local-gateway/base
@@ -0,0 +1,24 @@
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: kubeflow-ingress
namespace: istio-system
annotations:
nginx.ingress.kubernetes.io/rewrite-target: /
cert-manager.io/cluster-issuer: "kubeflow-self-signing-issuer"
spec:
tls:
- hosts:
- $HOST
secretName: kubeflow-tls
rules:
- host: $HOST
http:
paths:
- path: /
pathType: Prefix
backend:
service:
name: istio-ingressgateway
port:
number: 80
@@ -0,0 +1,42 @@
# Kubeflow Core Applications Kustomization
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
# Kubeflow Pipelines
- ../../apps/pipeline/upstream/env/cert-manager/platform-agnostic-multi-user
# Katib
- ../../apps/katib/upstream/installs/katib-with-kubeflow
# Central Dashboard
- ../../apps/centraldashboard/overlays/oauth2-proxy
# Admission Webhook
- ../../apps/admission-webhook/upstream/overlays/cert-manager
# Jupyter Web App
- ../../apps/jupyter/jupyter-web-app/upstream/overlays/istio
# Notebook Controller
- ../../apps/jupyter/notebook-controller/upstream/overlays/kubeflow
# Profiles + KFAM
- ../../apps/profiles/upstream/overlays/kubeflow
# PVC Viewer
- ../../apps/pvcviewer-controller/upstream/base
# Volumes Web App
- ../../apps/volumes-web-app/upstream/overlays/istio
# Tensorboards Controller
- ../../apps/tensorboard/tensorboard-controller/upstream/overlays/kubeflow
# Tensorboard Web App
- ../../apps/tensorboard/tensorboards-web-app/upstream/overlays/istio
# Training Operator
- ../../apps/training-operator/upstream/overlays/kubeflow
# User namespace
- ../../common/user-namespace/base
# KServe
- ../../apps/kserve/kserve
- ../../apps/kserve/models-web-app/overlays/kubeflow
# Spark Operator
- ../../apps/spark/spark-operator/overlays/kubeflow
# Ingress
- ingress.yaml
# dip
- cluster-policy.yaml
- cluster-role.yaml
- cluster-rolebinding.yaml
@@ -0,0 +1,19 @@
# Kubeflow Dependencies Kustomization
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
# Cert-Manager
- ../../common/cert-manager/kubeflow-issuer/base
# OAuth2-Proxy
- ../../common/oauth2-proxy/overlays/m2m-keycloak
# KNative
- ../../common/knative/knative-serving/overlays/gateways
# Kubeflow namespace
- ../../common/kubeflow-namespace/base
# NetworkPolicies
- ../../common/networkpolicies/base
# Kubeflow Roles
- ../../common/kubeflow-roles/base
# Kubeflow Istio Resources
- ../../common/istio-1-24/kubeflow-istio-resources/base