상황에 따라 추론의 요청처를 바꾸는 패턴입니다.
- Python 3.8 이상
- Docker
- Kubernetes 클러스터 또는 minikube
이 프로그램은 Kubernetes 클러스터 또는 minikube 가 필요합니다. Kubernetes 클러스터는 독자적으로 구축하거나, 각 클라우드 매니지드 서비스（GCP GKE、AWS EKS、MS Azure AKS 등）를 이용해 주십시오. GCP GKE 클러스터로 가동을 확인했습니다.
- 현재 디렉토리
$ pwd
~/ml-system-in-actions/chapter6_operation_management/condition_based_pattern- Docker 이미지 빌드
$ make build_all
# 실행 커맨드
# docker build \
# -t shibui/ml-system-in-actions:condition_based_pattern_proxy_0.0.1 \
# -f ./Dockerfile.proxy \
# .
# docker build \
# -t shibui/ml-system-in-actions:condition_based_pattern_imagenet_mobilenet_v2_0.0.1 \
# -f ./imagenet_mobilenet_v2/Dockerfile \
# .
# docker build \
# -t shibui/ml-system-in-actions:condition_based_pattern_plant_0.0.1 \
# -f ./plant/Dockerfile \
# .
# docker build \
# -t shibui/ml-system-in-actions:condition_based_pattern_client_0.0.1 \
# -f Dockerfile.client \
# .- Kubernetes 로 서비스 기동
$ make deploy
# 실행 커맨드
# istioctl install -y
# kubectl apply -f manifests/namespace.yml
# kubectl apply -f manifests/
# 가동확인
$ kubectl -n condition-based-serving get all
# 출력
# NAME READY STATUS RESTARTS AGE
# pod/client 2/2 Running 0 59s
# pod/mobilenet-v2-c9dc95c89-66dn9 2/2 Running 0 58s
# pod/mobilenet-v2-c9dc95c89-985sq 2/2 Running 0 58s
# pod/mobilenet-v2-c9dc95c89-qs5ht 2/2 Running 0 58s
# pod/mobilenet-v2-proxy-59d9f899c-7bpbb 1/2 Running 0 57s
# pod/mobilenet-v2-proxy-59d9f899c-blwz9 2/2 Running 0 57s
# pod/mobilenet-v2-proxy-59d9f899c-c2mlz 2/2 Running 0 57s
# pod/plant-9766f44d7-d9wwb 2/2 Running 0 58s
# pod/plant-9766f44d7-jn6m2 2/2 Running 0 58s
# pod/plant-9766f44d7-lnn48 2/2 Running 0 58s
# pod/plant-proxy-558877db5c-d8wvz 2/2 Running 0 57s
# pod/plant-proxy-558877db5c-fwtsh 1/2 Running 0 57s
# pod/plant-proxy-558877db5c-qpxmz 2/2 Running 0 57s
# NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
# service/mobilenet-v2 ClusterIP 10.4.8.216 <none> 8500/TCP,8501/TCP 58s
# service/plant ClusterIP 10.4.9.128 <none> 9500/TCP,9501/TCP 57s
# service/proxy ClusterIP 10.4.11.158 <none> 8000/TCP 57s
# NAME READY UP-TO-DATE AVAILABLE AGE
# deployment.apps/mobilenet-v2 3/3 3 3 59s
# deployment.apps/mobilenet-v2-proxy 2/3 3 2 58s
# deployment.apps/plant 3/3 3 3 59s
# deployment.apps/plant-proxy 2/3 3 2 58s
# NAME DESIRED CURRENT READY AGE
# replicaset.apps/mobilenet-v2-c9dc95c89 3 3 3 59s
# replicaset.apps/mobilenet-v2-proxy-59d9f899c 3 3 2 58s
# replicaset.apps/plant-9766f44d7 3 3 3 59s
# replicaset.apps/plant-proxy-558877db5c 3 3 2 58s- 기동한 API 에 요청
# 클라이언트에 접속
$ kubectl -n condition-based-serving exec -it pod/client bash
# 헬스 체크
$ curl proxy.condition-based-serving.svc.cluster.local:8000/health
# 출력
# {
# "health":"ok"
# }
# 메타 데이터
$ curl proxy.condition-based-serving.svc.cluster.local:8000/metadata
# 출력
# {
# "data_type": "str",
# "data_structure": "(1,1)",
# "data_sample": "base64 encoded image file",
# "prediction_type": "float32",
# "prediction_structure": "(1,1001)",
# "prediction_sample": "[0.07093159, 0.01558308, 0.01348537, ...]"
# }
# 라벨 목록
$ curl proxy.condition-based-serving.svc.cluster.local:8000/label
# 출력
# [
# "background",
# "tench",
# "goldfish",
# ...
# "bolete",
# "ear",
# "toilet tissue"
# ]
# 테스트 데이터로 추론 요청
$ curl proxy.condition-based-serving.svc.cluster.local:8000/predict/test
# 출력
# "Persian cat"
# 고양이 이미지를 ImageNet 추론기로 요청
$ (echo -n '{"image_data": "'; base64 cat.jpg; echo '"}') | \
curl \
-X POST \
-H "Content-Type: application/json" \
-d @- \
proxy.condition-based-serving.svc.cluster.local:8000/predict
# 출력
# "Persian cat"
# 붓꽃 이미지를 식물 추론기로 요청
$ (echo -n '{"image_data": "'; base64 iris.jpg; echo '"}') | \
curl \
-X POST \
-H "Content-Type: application/json" \
-H "target: mountain" \
-d @- \
proxy.condition-based-serving.svc.cluster.local:8000/predict
# 출력
# "Iris versicolor"
# 고양이 이미지를 식물 추론기로 요청을 보내면, background 로 분류됨
$ (echo -n '{"image_data": "'; base64 cat.jpg; echo '"}') | \
curl \
-X POST \
-H "Content-Type: application/json" \
-H "target: mountain" \
-d @- \
proxy.condition-based-serving.svc.cluster.local:8000/predict
# 출력
# "background"
# 붓꽃 이미지를 ImageNet 추론기로 요청을 보내면, bee 로 분류됨
$ (echo -n '{"image_data": "'; base64 iris.jpg; echo '"}') | \
curl \
-X POST \
-H "Content-Type: application/json" \
-d @- \
proxy.condition-based-serving.svc.cluster.local:8000/predict
# 출력
# "bee"- Kubernetes 에서 서비스 삭제
$ kubectl delete ns condition-based-serving
$ istioctl x uninstall --purge -y
$ kubectl delete ns istio-system