docs: change_for_airflow_provider_demo_of_integration_folder (#3467)

docs/en/integration/deploy_integration/airflow_provider_demo.md

@@ -0,0 +1,145 @@
+# Airflow 
+We provide [Airflow OpenMLDB Provider](https://github.com/4paradigm/OpenMLDB/tree/main/extensions/airflow-provider-openmldb), which facilitates the integration of OpenMLDB with Airflow DAG.
+
+This specific case will undergo training and execution with Airflow's [TalkingData](https://chat.openai.com/talkingdata_demo).
+
+## TalkingData DAG
+
+To implement this workflow in Airflow, a DAG (Directed Acyclic Graph) file needs to be written. Here we use an example DAG file in [example_openmldb_complex.py](https://github.com/4paradigm/OpenMLDB/blob/main/extensions/airflow-provider-openmldb/openmldb_provider/example_dags/example_openmldb_complex.py).
+
+![airflow dag](images/airflow_dag.png)
+
+The diagram above illustrates the work process in DAG. It begins by creating a table, followed by offline data loading, feature extraction, and model training. If the model trained performs well (AUC >= 99.0), the workflow proceeds to execute deploy SQL and model serving online. Otherwise, a failure report is generated.
+
+In the following demonstration, you can directly import this DAG and run in Airflow.
+
+
+## Demonstration
+
+We import the above DAG to perform feature computation and deployment for the TalkingData Demo, then perform real-time inference using the predict server of TalkingData Demo.
+
+### Preparation
+
+#### Download DAG
+
+Along with the DAG files, training scripts are also required. For convenience, we provide the [code package](https://openmldb.ai/download/airflow_demo/airflow_demo_files.tar.gz) for direct download. If you prefer to use the latest version, you can obtain it from [github example_dags](https://github.com/4paradigm/OpenMLDB/tree/main/extensions/airflow-provider-openmldb/openmldb_provider/example_dags).
+
+```
+wget https://openmldb.ai/download/airflow_demo/airflow_demo_files.tar.gz
+tar zxf airflow_demo_files.tar.gz
+ls airflow_demo_files
+```
+#### Start Docker Image
+
+For smooth function, we recommend starting OpenMLDB using the docker image and installing Airflow within the docker container.
+
+Since Airflow Web requires an external port for login, the container's port must be exposed. Then map the downloaded file from the previous step to the `/work/airflow/dags` directory. This step is crucial for Airflow to load the DAGs from this folder correctly.
+
+```
+docker run -p 8080:8080 -v `pwd`/airflow_demo_files:/work/airflow_demo_files -it 4pdosc/openmldb:0.8.0 bash
+```
+
+#### Download and Install Airflow and Airflow OpenMLDB Provider
+In the docker container, execute:
+```
+pip3 install airflow-provider-openmldb
+```
+Airflow will be downloaded as a dependency.
+
+#### Source Data and DAG Preparation
+Copy the sample data file, named `/tmp/train_sample.csv`, to the tmp directory. Airflow DAG files and training scripts used in the DAG must also be copied to the Airflow directory.
+
+```
+cp /work/airflow_demo_files/train_sample.csv /tmp/
+mkdir -p /work/airflow/dags
+cp /work/airflow_demo_files/example_openmldb_complex.py /work/airflow_demo_files/xgboost_train_sample.py /work/airflow/dags
+```
+
+### Step 1: Start OpenMLDB and Airflow
+The command provided below will initiate the OpenMLDB cluster, enabling support for predict server and Airflow standalone.
+```
+/work/init.sh
+python3 /work/airflow_demo_files/predict_server.py --no-init > predict.log 2>&1 &
+export AIRFLOW_HOME=/work/airflow
+cd $AIRFLOW_HOME
+airflow standalone
+```
+
+Airflow standalone will show username and password as shown below.
+
+![airflow login](images/airflow_login.png)
+
+In Airflow Web interface at `http://localhost:8080`, enter username and password.
+
+```{caution}
+`airflow standalone` is a front-end program that exits with Airflow. You can exit Airflow after DAG completion to run [Step 3-Testing](#3-Testing), or place the Airflow process in the background.
+```
+
+### Step 2: Running DAG
+
+To check the status of the DAG "example_openmldb_complex" in Airflow Web, click on the DAG and select the `Code` tab, as shown below.
+
+![dag home](images/dag_home.png)
+
+In this code, you will notice the usage of `openmldb_conn_id`, as depicted in the following figure. The DAG doesn't directly employ the address of OpenMLDB; instead, it uses a connection, so you need to create a new connection with the same name.
+
+![dag code](images/dag_code.png)
+
+#### Create Connection
+Click on connections in the Admin tab.
+![connection](images/connection.png)
+
+Add the connection.
+![add connection](images/add_connection.png)
+
+The Airflow OpenMLDB Provider is linked to the OpenMLDB API Server. Therefore, you should provide the address of the OpenMLDB API Server in this configuration, rather than the Zookeeper address.
+
+![connection settings](images/connection_settings.png)
+
+The completed connection is shown in the figure below.
+![display](images/connection_display.png)
+
+#### Running DAG
+Run the DAG to complete the training of the model, SQL deployment, and model deployment. A successful operation will yield results similar to the figure below.
+![dag run](images/dag_run.png)
+
+### Step 3: Test
+
+If Airflow is currently running in the foreground within the container, you may exit the process now. The upcoming tests will not be dependent on Airflow.
+
+#### Online Data Import
+The SQL and model deployment have been successfully executed in the Airflow DAG. However, there is currently no data in the online storage, necessitating an online data import.
+
+```
+curl -X POST http://127.0.0.1:9080/dbs/example_db -d'{"mode":"online", "sql":"load data infile \"file:///tmp/train_sample.csv\" into table example_table options(mode=\"append\");"}'
+```
+
+This import process is asynchronous, but since the data volume is small, it will be completed quickly. You can monitor the status of the import operations by using the `SHOW JOBS` command.
+```
+curl -X POST http://127.0.0.1:9080/dbs/example_db -d'{"mode":"online", "sql":"show jobs"}'
+```
+
+#### Prediction
+Execute the prediction script to make a prediction using the newly deployed SQL and model.
+```
+python3 /work/airflow_demo_files/predict.py
+```
+The result is as shown.
+![result](images/airflow_test_result.png)
+
+
+### Non-Interactive Testing
+
+Check if DAG has been successfully loaded:
+```
+airflow dags list | grep openmldb
+```
+Add required connection:
+```
+airflow connections add openmldb_conn_id --conn-uri http://127.0.0.1:9080
+airflow connections list --conn-id openmldb_conn_id
+```
+DAG test:
+```
+airflow dags test example_openmldb_complex 2022-08-25
+```

docs/zh/integration/deploy_integration/airflow_provider_demo.md

@@ -17,9 +17,9 @@ DAG流程如上图所示，首先建表，然后进行离线数据导入与特
 
 我们导入上述的DAG完成TalkingData Demo中的特征计算与上线，并使用TalkingData Demo的predict server来进行上线后的实时推理测试。
 
-### 0 准备
+### 准备工作
 
-#### 0.1 下载DAG
+#### 下载DAG
 
 除了DAG文件，还需要训练的脚本，所以我们提供了[下载包](https://openmldb.ai/download/airflow_demo/airflow_demo_files.tar.gz)，可以直接下载。如果想要使用最新版本，请在[github example_dags](https://github.com/4paradigm/OpenMLDB/tree/main/extensions/airflow-provider-openmldb/openmldb_provider/example_dags)中获取。
 
@@ -28,7 +28,7 @@ wget https://openmldb.ai/download/airflow_demo/airflow_demo_files.tar.gz
 tar zxf airflow_demo_files.tar.gz
 ls airflow_demo_files
 ```
-#### 0.2 启动镜像
+#### 启动镜像
 
 我们推荐使用docker镜像直接启动OpenMLDB，并在docker内部安装启动Airflow。
 
@@ -38,22 +38,22 @@ ls airflow_demo_files
 docker run -p 8080:8080 -v `pwd`/airflow_demo_files:/work/airflow_demo_files -it 4pdosc/openmldb:0.8.3 bash
 ```
 
-#### 0.3 下载安装Airflow与Airflow OpenMLDB Provider
+#### 下载安装Airflow与Airflow OpenMLDB Provider
 在docker容器中，执行：
 ```
 pip3 install airflow-provider-openmldb
 ```
 由于airflow-provider-openmldb依赖airflow，所以会一起下载。
 
-#### 0.4 源数据与DAG准备
+#### 源数据与DAG准备
 由于在DAG中导入数据用的文件为`/tmp/train_sample.csv`，所以我们需要将sample数据文件拷贝到tmp目录。Airflow 的DAG文件和DAG中使用的训练脚本也需要拷贝到airflow目录中。
 ```
 cp /work/airflow_demo_files/train_sample.csv /tmp/
 mkdir -p /work/airflow/dags
 cp /work/airflow_demo_files/example_openmldb_complex.py /work/airflow_demo_files/xgboost_train_sample.py /work/airflow/dags
 ```
 
-### 1 启动OpenMLDB与Airflow
+### 步骤1：启动OpenMLDB与Airflow
 以下命令将启动OpenMLDB cluster，支持上线并测试的predict server，与Airflow standalone。
 ```
 /work/init.sh
@@ -73,7 +73,7 @@ Airflow standalone运行输出将提示登录用户名和密码，如下图所
 `airflow standalone`为前台程序，退出即airflow退出。你可以在dag运行完成后再退出airflow进行[第三步————测试](#3-测试)，或者将airflow进程放入后台。
 ```
 
-### 2 运行DAG
+### 步骤2：运行DAG
 在Airflow Web中点击DAG example_openmldb_complex，可以点击`Code`查看DAG的详情，见下图。
 
 ![dag home](images/dag_home.png)
@@ -82,7 +82,7 @@ Airflow standalone运行输出将提示登录用户名和密码，如下图所
 
 ![dag code](images/dag_code.png)
 
-#### 2.1 创建connection
+#### 创建connection
 在管理界面中点击connection。
 ![connection](images/connection.png)
 
@@ -96,15 +96,15 @@ Airflow OpenMLDB Provider是连接OpenMLDB Api Server的，所以此处配置中
 创建完成后的connection如下图所示。
 ![display](images/connection_display.png)
 
-#### 2.2 运行DAG
+#### 运行DAG
 运行dag，即完成一次训练模型、sql部署与模型部署。成功运行的结果，类似下图。
 ![dag run](images/dag_run.png)
 
-### 3 测试
+### 步骤3：测试
 
 Airflow如果在容器中是前台运行的，现在可以退出，以下测试将不依赖airflow。
 
-#### 3.1 在线导入
+#### 在线导入
 Airflow DAG中完成了SQL和模型的上线。但在线存储中还没有数据，所以我们需要做一次在线数据导入。
 ```
 curl -X POST http://127.0.0.1:9080/dbs/example_db -d'{"mode":"online", "sql":"load data infile \"file:///tmp/train_sample.csv\" into table example_table options(mode=\"append\");"}'
@@ -115,7 +115,7 @@ curl -X POST http://127.0.0.1:9080/dbs/example_db -d'{"mode":"online", "sql":"lo
 curl -X POST http://127.0.0.1:9080/dbs/example_db -d'{"mode":"online", "sql":"show jobs"}'
 ```
 
-#### 3.2 预测
+#### 预测
 执行预测脚本，进行一次预测，预测将使用新部署好的sql与模型。
 ```
 python3 /work/airflow_demo_files/predict.py