Docker部署mlflow服务器

部署MySQL

1. docker pull mysql:latest

2. docker run
   -p 3306:3306
   –name ml-mysql
   -v ~/mysql/data:/var/lib/mysql
   -e MYSQL_ROOT_PASSWORD=669988
   -d mysql

3. 在mysql里创建一个叫mlflow的库

create database if not exists `mlflow`;

部署Minio

1. docker pull minio

2. docker run -d -p 9000:9000 -p 9090:9090
   –name minio
   -e “MINIO_ACCESS_KEY=minio”
   -e “MINIO_SECRET_KEY=minio123”
   -v /opt/minio/data:/data
   -v /opt/minio/config:/root/.minio
   minio/minio server /data
   –console-address “:9000” --address “:9090”
   // the port 9000 represents minio console port
   // the port 9090 represents minio api port

3. 在minio的web ui上创建一个叫mlflow的bucket

4. 网上教程一般都有这步，但我从来没成功过
   export AWS_ACCESS_KEY_ID=minio
   export AWS_SECRET_ACCESS_KEY=minio123
   export MLFLOW_S3_ENDPOINT_URL=http://localhost:9000

装python环境

1. 创建一个新环境

conda create -n mltest python==3.8

2. pip install mlflow mysqlclient/pymysql boto3

启动MLflow

1. mlflow server
   –backend-store-uri mysql+pymysql://root:669988@localhost/mlflow
   –host 0.0.0.0 -p 5002
   –default-artifact-root s3://mlflow

MLflow访问Minio

1. 把这些加在代码里
   os.environ[‘MLFLOW_S3_ENDPOINT_URL’] = ‘http://888.88.88.888:9090/’
   os.environ[‘AWS_ACCESS_KEY_ID’] = ‘minio’
   os.environ[‘AWS_SECRET_ACCESS_KEY’] = ‘minio123’

mlflow使用

import boto3
import mlflow
import mlflow.sklearn
#设置远程路径
mlflow.set_tracking_uri("http://888.88.88.888:5002")
mlflow.set_experiment("my-experiment")

with mlflow.start_run():
    # 记录模型param
	mlflow.log_param("alpha", alpha)
	mlflow.log_params(dict) # 多个param传个dict进去
	# 记录模型metric
    mlflow.log_metric("r2", r2)
    mlflow.log_metrics(dict) # metrics同理
    # 记录artifacts
    mlflow.sklearn.log_model(lr, "model")
    # Log the sklearn model and register as version 1
    mlflow.sklearn.log_model(
        sk_model=lr,
        artifact_path="sklearn-model",
        registered_model_name="sk-learn-random-forest-reg-model",
    )
    mlflow.end_run()