clean-up e persistencia na nuvem

This commit is contained in:
2026-05-08 08:29:08 -03:00
parent f58cdb63e2
commit 1b13e119a9
3 changed files with 788 additions and 37 deletions

Binary file not shown.

View File

@@ -0,0 +1,442 @@
apiVersion: v1
kind: Secret
metadata:
name: langfuse-external-services
namespace: ${K8S_NAMESPACE}
type: Opaque
stringData:
# PostgreSQL OCI
DATABASE_URL: "postgresql://langfuse:${DB_PASSWORD}@10.0.10.15:5432/langfuse"
# Redis OCI
REDIS_HOST: "10.0.20.10"
REDIS_PORT: "6379"
REDIS_AUTH: "${REDIS_AUTH}"
# ClickHouse
CLICKHOUSE_PASSWORD: "${CLICKHOUSE_PASSWORD}"
# Langfuse
NEXTAUTH_SECRET: "${NEXTAUTH_SECRET}"
SALT: "${SALT}"
ENCRYPTION_KEY: "${ENCRYPTION_KEY}"
# OCI Object Storage
S3_ACCESS_KEY: "${S3_ACCESS_KEY}"
S3_SECRET_KEY: "${S3_SECRET_KEY}"
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: ${APP_NAME}-clickhouse
namespace: ${K8S_NAMESPACE}
spec:
serviceName: ${APP_NAME}-clickhouse
replicas: 1
selector:
matchLabels:
app: ${APP_NAME}-clickhouse
template:
metadata:
labels:
app: ${APP_NAME}-clickhouse
spec:
imagePullSecrets:
- name: ocirsecret
containers:
- name: clickhouse
image: ${IMAGE_REPOSITORY}/clickhouse:${CLICKHOUSE_TAG}
ports:
- containerPort: 8123
- containerPort: 9000
resources:
requests:
cpu: "2"
memory: "8Gi"
limits:
cpu: "4"
memory: "16Gi"
volumeMounts:
- name: data
mountPath: /var/lib/clickhouse
env:
- name: CLICKHOUSE_DB
value: "default"
- name: CLICKHOUSE_USER
value: "default"
- name: CLICKHOUSE_PASSWORD
value: "${CLICKHOUSE_PASSWORD}"
- name: CLICKHOUSE_DEFAULT_ACCESS_MANAGEMENT
value: "1"
volumeClaimTemplates:
- metadata:
name: data
spec:
accessModes: ["ReadWriteOnce"]
storageClassName: oci-bv-high-performance
resources:
requests:
storage: 200Gi
---
apiVersion: v1
kind: Service
metadata:
name: ${APP_NAME}-clickhouse
namespace: ${K8S_NAMESPACE}
spec:
selector:
app: ${APP_NAME}-clickhouse
ports:
- name: http
port: 8123
- name: native
port: 9000
---
# ===== LANGFUSE APPLICATION =====
apiVersion: apps/v1
kind: Deployment
metadata:
name: ${APP_NAME}
namespace: ${K8S_NAMESPACE}
spec:
replicas: 1
selector:
matchLabels:
app: ${APP_NAME}
template:
metadata:
labels:
app: ${APP_NAME}
spec:
imagePullSecrets:
- name: ocirsecret
initContainers:
- name: wait-for-postgres
image: busybox
env:
- name: DB_HOST
value: "10.0.10.15"
command:
- sh
- -c
- |
until nc -z $DB_HOST 5432; do
echo "⏳ Waiting for PostgreSQL..."
sleep 2
done
- name: wait-for-redis
image: busybox
env:
- name: REDIS_HOST
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: REDIS_HOST
command:
- sh
- -c
- |
until nc -z $REDIS_HOST 6379; do
echo "⏳ Waiting for Redis..."
sleep 2
done
- name: wait-for-clickhouse
image: busybox
command:
- sh
- -c
- |
until nc -z ${APP_NAME}-clickhouse 9000; do
echo "⏳ Waiting for ClickHouse..."
sleep 2
done
containers:
- name: ${APP_NAME}
image: ${IMAGE_REPOSITORY}/langfuse:${LANGFUSE_TAG}
ports:
- containerPort: 3000
env:
- name: DATABASE_URL
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: DATABASE_URL
- name: NEXTAUTH_SECRET
value: "${NEXTAUTH_SECRET}"
- name: SALT
value: "${SALT}"
- name: ENCRYPTION_KEY
value: "${ENCRYPTION_KEY}"
- name: NEXTAUTH_URL
value: ${NEXTAUTH_URL}
- name: TELEMETRY_ENABLED
value: "false"
- name: CLICKHOUSE_URL
value: "http://${APP_NAME}-clickhouse:8123"
- name: CLICKHOUSE_USER
value: "default"
- name: CLICKHOUSE_PASSWORD
value: "${CLICKHOUSE_PASSWORD}"
- name: CLICKHOUSE_CLUSTER_ENABLED
value: "false"
- name: CLICKHOUSE_MIGRATION_URL
value: "clickhouse://${APP_NAME}-clickhouse:9000"
- name: REDIS_HOST
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: REDIS_HOST
- name: REDIS_AUTH
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: REDIS_AUTH
# ===== OCI OBJECT STORAGE (EVENTS) =====
- name: LANGFUSE_S3_EVENT_UPLOAD_BUCKET
value: "langfuse-events"
- name: LANGFUSE_S3_EVENT_UPLOAD_PREFIX
value: "events/"
- name: LANGFUSE_S3_EVENT_UPLOAD_REGION
value: "${OCI_REGION}"
- name: LANGFUSE_S3_EVENT_UPLOAD_ENDPOINT
value: "https://${OCI_NAMESPACE}.compat.objectstorage.${OCI_REGION}.oraclecloud.com"
- name: LANGFUSE_S3_EVENT_UPLOAD_ACCESS_KEY_ID
value: "${S3_ACCESS_KEY}"
- name: LANGFUSE_S3_EVENT_UPLOAD_SECRET_ACCESS_KEY
value: "${S3_SECRET_KEY}"
- name: LANGFUSE_S3_EVENT_UPLOAD_FORCE_PATH_STYLE
value: "true"
resources:
requests:
cpu: "50m"
memory: "2Gi"
limits:
cpu: "512m"
memory: "3Gi"
livenessProbe:
httpGet:
path: /api/public/health
port: 3000
initialDelaySeconds: 30
periodSeconds: 15
readinessProbe:
httpGet:
path: /api/public/health
port: 3000
initialDelaySeconds: 15
periodSeconds: 10
---
# ===== LANGFUSE WORKER =====
apiVersion: apps/v1
kind: Deployment
metadata:
name: ${APP_NAME}-worker
namespace: ${K8S_NAMESPACE}
spec:
replicas: 1
selector:
matchLabels:
app: ${APP_NAME}-worker
template:
metadata:
labels:
app: ${APP_NAME}-worker
spec:
imagePullSecrets:
- name: ocirsecret
initContainers:
- name: wait-for-postgres
image: busybox
command:
- sh
- -c
- |
until nc -z $DB_HOST 5432; do
echo "⏳ Waiting for PostgreSQL..."
sleep 2
done
- name: wait-for-redis
image: busybox
command:
- sh
- -c
- |
until nc -z $REDIS_HOST 6379; do
echo "⏳ Waiting for Redis..."
sleep 2
done
- name: wait-for-clickhouse
image: busybox
command:
- sh
- -c
- |
until nc -z ${APP_NAME}-clickhouse 9000; do
echo "⏳ Waiting for ClickHouse..."
sleep 2
done
containers:
- name: worker
image: ${IMAGE_REPOSITORY}/langfuse-worker:${LANGFUSE_WORKER_TAG}
env:
- name: DATABASE_URL
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: DATABASE_URL
- name: ENCRYPTION_KEY
value: "${ENCRYPTION_KEY}"
- name: CLICKHOUSE_URL
value: "http://${APP_NAME}-clickhouse:8123"
- name: CLICKHOUSE_USER
value: "default"
- name: CLICKHOUSE_PASSWORD
value: "${CLICKHOUSE_PASSWORD}"
- name: CLICKHOUSE_CLUSTER_ENABLED
value: "false"
- name: CLICKHOUSE_MIGRATION_URL
value: "clickhouse://${APP_NAME}-clickhouse:9000"
- name: REDIS_HOST
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: REDIS_HOST
- name: REDIS_PORT
value: "6379"
- name: REDIS_AUTH
valueFrom:
secretKeyRef:
name: langfuse-external-services
key: REDIS_AUTH
# ===== OCI OBJECT STORAGE (EVENTS) =====
- name: LANGFUSE_S3_EVENT_UPLOAD_BUCKET
value: "langfuse-events"
- name: LANGFUSE_S3_EVENT_UPLOAD_PREFIX
value: "events/"
- name: LANGFUSE_S3_EVENT_UPLOAD_REGION
value: "${OCI_REGION}"
- name: LANGFUSE_S3_EVENT_UPLOAD_ENDPOINT
value: "https://${OCI_NAMESPACE}.compat.objectstorage.${OCI_REGION}.oraclecloud.com"
- name: LANGFUSE_S3_EVENT_UPLOAD_ACCESS_KEY_ID
value: "${S3_ACCESS_KEY}"
- name: LANGFUSE_S3_EVENT_UPLOAD_SECRET_ACCESS_KEY
value: "${S3_SECRET_KEY}"
- name: LANGFUSE_S3_EVENT_UPLOAD_FORCE_PATH_STYLE
value: "true"
resources:
requests:
cpu: "50m"
memory: "2Gi"
limits:
cpu: "512m"
memory: "3Gi"
---
# ===== LANGFUSE SERVICE =====
apiVersion: v1
kind: Service
metadata:
name: langfuse-web
namespace: ${K8S_NAMESPACE}
annotations:
service.beta.kubernetes.io/oci-load-balancer-internal: "true"
oci.oraclecloud.com/healthcheck-path: "/api/public/health"
oci.oraclecloud.com/healthcheck-port: "3000"
spec:
type: LoadBalancer
selector:
app: ${APP_NAME}
ports:
- protocol: TCP
port: 80
targetPort: 3000
---
# ===== LANGFUSE HPA - HORIZONTAL POD AUTOSCALER =====
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: ${APP_NAME}-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: ${APP_NAME}
minReplicas: 2
maxReplicas: 10
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 70
---
# ===== WORKER - AUTOSCALER =====
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: langfuse-worker-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: ${APP_NAME}-worker
minReplicas: 2
maxReplicas: 10
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 70
---
apiVersion: v1
kind: Service
metadata:
name: ${APP_NAME}-lb
namespace: ${K8S_NAMESPACE}
annotations:
service.beta.kubernetes.io/oci-load-balancer-shape: "flexible"
oci.oraclecloud.com/healthcheck-path: "/api/public/health"
spec:
type: LoadBalancer
selector:
app: ${APP_NAME}
ports:
- port: 80
targetPort: 3000