Files
mlflow--mlflow/charts/example-mlflow-charts.yaml
2026-07-13 13:22:34 +08:00

98 lines
2.4 KiB
YAML

# Example values for the MLflow Helm chart (SQLite + local storage).
# Usage:
# helm upgrade --install mlflow ./charts --namespace mlflow --create-namespace -f charts/example-mlflow-charts.yaml
image:
repository: ghcr.io/mlflow/mlflow
tag: "v3.11.1-full"
replicaCount: 1
server:
value_options:
host: "0.0.0.0"
port: 5000
workers: 4
# Uncomment to enable the basic-auth plugin:
# app_name: "basic-auth"
# Bare flags rendered as --flag. Uncomment as needed.
flag_options: []
# - no_serve_artifacts
mlflow:
backendStoreUri: "sqlite:////mlflow/mlflow.db"
artifactsDestination: "/mlflow/artifacts"
# --- For production: use a real database and object store instead ---
# Option A: inline URI (password visible in values — development only).
# Option B: reference a Secret (recommended for production).
#
# Create the Secret first:
# kubectl create secret generic mlflow-db-secret \
# --from-literal=uri="postgresql://user:password@postgres:5432/mlflow"
# backendStoreUri: "postgresql://user:password@postgres:5432/mlflow"
# backendStoreUriFrom:
# secretKeyRef:
# name: mlflow-db-secret
# key: uri
# artifactsDestination: "s3://my-bucket/mlflow"
storage:
enabled: true
# StorageClass to use. Set to match your cluster
# (e.g. "gp2" for standard EKS, "standard" for Kind/minikube).
storageClassName: "gp2"
# S3 credentials (or other artifact store credentials):
# kubectl create secret generic s3-credentials \
# --from-literal=access-key-id=<key> \
# --from-literal=secret-access-key=<secret>
# env:
# - name: AWS_ACCESS_KEY_ID
# valueFrom:
# secretKeyRef:
# name: s3-credentials
# key: access-key-id
# - name: AWS_SECRET_ACCESS_KEY
# valueFrom:
# secretKeyRef:
# name: s3-credentials
# key: secret-access-key
service:
type: ClusterIP
# Expose via Ingress:
# ingress:
# enabled: true
# className: nginx
# hosts:
# - host: mlflow.example.com
# paths:
# - path: /
# pathType: Prefix
# Prometheus metrics:
metrics:
enabled: false
# Prometheus ServiceMonitor (requires Prometheus Operator):
# serviceMonitor:
# enabled: true
# Garbage collection (removes permanently soft-deleted resources):
garbageCollection:
enabled: true
schedule: "0 2 * * 0"
olderThan: "30d"
resources:
requests:
cpu: 500m
memory: 512Mi
limits:
cpu: 2000m
memory: 2Gi