Muokkaa

Jaa

Submit and manage jobs on an Apache Spark™ cluster in HDInsight on AKS

  • Artikkeli

Note

We will retire Azure HDInsight on AKS on January 31, 2025. Before January 31, 2025, you will need to migrate your workloads to Microsoft Fabric or an equivalent Azure product to avoid abrupt termination of your workloads. The remaining clusters on your subscription will be stopped and removed from the host.

Only basic support will be available until the retirement date.

Important

This feature is currently in preview. The Supplemental Terms of Use for Microsoft Azure Previews include more legal terms that apply to Azure features that are in beta, in preview, or otherwise not yet released into general availability. For information about this specific preview, see Azure HDInsight on AKS preview information. For questions or feature suggestions, please submit a request on AskHDInsight with the details and follow us for more updates on Azure HDInsight Community.

Once the cluster is created, user can use various interfaces to submit and manage jobs by

  • using Jupyter
  • using Zeppelin
  • using ssh (spark-submit)

Using Jupyter

Prerequisites

An Apache Spark™ cluster on HDInsight on AKS. For more information, see Create an Apache Spark cluster.

Jupyter Notebook is an interactive notebook environment that supports various programming languages.

Create a Jupyter Notebook

  1. Navigate to the Apache Spark™ cluster page and open the Overview tab. Click on Jupyter, it asks you to authenticate and open the Jupyter web page.

    Screenshot of how to select Jupyter notebook.

  2. From the Jupyter web page, Select New > PySpark to create a notebook.

    Screenshot of new PySpark page.

    A new notebook created and opened with the name Untitled(Untitled.ipynb).

    Note

    By using the PySpark or the Python 3 kernel to create a notebook, the spark session is automatically created for you when you run the first code cell. You do not need to explicitly create the session.

  3. Paste the following code in an empty cell of the Jupyter Notebook, and then press SHIFT + ENTER to run the code. See here for more controls on Jupyter.

    Screenshot of PySpark page with contents. 

    %matplotlib inline
import pandas as pd
import matplotlib.pyplot as plt
data1 = [22,40,10,50,70]
s1 = pd.Series(data1)   #One-dimensional ndarray with axis labels (including time series).

data2 = data1
index = ['John','sam','anna','smith','ben']
s2 = pd.Series(data2,index=index)

data3 = {'John':22, 'sam':40, 'anna':10,'smith':50,'ben':70}
s3 = pd.Series(data3)

s3['jp'] = 32     #insert a new row
s3['John'] = 88

names = ['John','sam','anna','smith','ben']
ages = [10,40,50,48,70]
name_series = pd.Series(names)
age_series = pd.Series(ages)

data_dict = {'name':name_series, 'age':age_series}
dframe = pd.DataFrame(data_dict)   
#create a pandas DataFrame from dictionary

dframe['age_plus_five'] = dframe['age'] + 5   
#create a new column
dframe.pop('age_plus_five')
#dframe.pop('age')

salary = [1000,6000,4000,8000,10000]
salary_series = pd.Series(salary)
new_data_dict = {'name':name_series, 'age':age_series,'salary':salary_series}
new_dframe = pd.DataFrame(new_data_dict)
new_dframe['average_salary'] = new_dframe['age']*90

new_dframe.index = new_dframe['name']
print(new_dframe.loc['sam'])

  4. Plot a graph with Salary and age as the X and Y axes

  5. In the same notebook, paste the following code in an empty cell of the Jupyter Notebook, and then press SHIFT + ENTER to run the code.

    %matplotlib inline
import pandas as pd
import matplotlib.pyplot as plt

plt.plot(age_series,salary_series)
plt.show()

    Screenshot of graph output.

Save the Notebook

  1. From the notebook menu bar, navigate to File > Save and Checkpoint.

  2. Shut down the notebook to release the cluster resources: from the notebook menu bar, navigate to File > Close and Halt. You can also run any of the notebooks under the examples folder.

    Screenshot of how to save the note books.

Using Apache Zeppelin notebooks

Apache Spark clusters in HDInsight on AKS include Apache Zeppelin notebooks. Use the notebooks to run Apache Spark jobs. In this article, you learn how to use the Zeppelin notebook on an HDInsight on AKS cluster.

Prerequisites

An Apache Spark cluster on HDInsight on AKS. For instructions, see Create an Apache Spark cluster.

Launch an Apache Zeppelin notebook

  1. Navigate to the Apache Spark cluster Overview page and select Zeppelin notebook from Cluster dashboards. It prompts to authenticate and open the Zeppelin page.

    Screenshot of how to select Zeppelin.

  2. Create a new notebook. From the header pane, navigate to Notebook > Create new note. Ensure the notebook header shows a connected status. It denotes a green dot in the top-right corner.

    Screenshot of how to create zeppelin notebook.

  3. Run the following code in Zeppelin Notebook:

    %livy.pyspark
import pandas as pd
import matplotlib.pyplot as plt
data1 = [22,40,10,50,70]
s1 = pd.Series(data1)   #One-dimensional ndarray with axis labels (including time series).

data2 = data1
index = ['John','sam','anna','smith','ben']
s2 = pd.Series(data2,index=index)

data3 = {'John':22, 'sam':40, 'anna':10,'smith':50,'ben':70}
s3 = pd.Series(data3)

s3['jp'] = 32     #insert a new row

s3['John'] = 88

names = ['John','sam','anna','smith','ben']
ages = [10,40,50,48,70]
name_series = pd.Series(names)
age_series = pd.Series(ages)

data_dict = {'name':name_series, 'age':age_series}
dframe = pd.DataFrame(data_dict)   #create a pandas DataFrame from dictionary

dframe['age_plus_five'] = dframe['age'] + 5   #create a new column
dframe.pop('age_plus_five')
#dframe.pop('age')

salary = [1000,6000,4000,8000,10000]
salary_series = pd.Series(salary)
new_data_dict = {'name':name_series, 'age':age_series,'salary':salary_series}
new_dframe = pd.DataFrame(new_data_dict)
new_dframe['average_salary'] = new_dframe['age']*90

new_dframe.index = new_dframe['name']
print(new_dframe.loc['sam'])

  4. Select the Play button for the paragraph to run the snippet. The status on the right-corner of the paragraph should progress from READY, PENDING, RUNNING to FINISHED. The output shows up at the bottom of the same paragraph. The screenshot looks like the following image:

    Screenshot of how to run Zeppelin notebook.

    Output:

    Screenshot of Zeppelin notebook output.

Using Spark submit jobs

  1. Create a file using the following command `#vim samplefile.py'

  2. This command opens the vim file

  3. Paste the following code into the vim file

    import pandas as pd
import matplotlib.pyplot as plt

From pyspark.sql import SparkSession
Spark = SparkSession.builder.master('yarn').appName('SparkSampleCode').getOrCreate()
# Initialize spark context
data1 = [22,40,10,50,70]
s1 = pd.Series(data1)   #One-dimensional ndarray with axis labels (including time series).

data2 = data1
index = ['John','sam','anna','smith','ben']
s2 = pd.Series(data2,index=index)

data3 = {'John':22, 'sam':40, 'anna':10,'smith':50,'ben':70}
 s3 = pd.Series(data3)

s3['jp'] = 32     #insert a new row

s3['John'] = 88

names = ['John','sam','anna','smith','ben']
ages = [10,40,50,48,70]
name_series = pd.Series(names)
age_series = pd.Series(ages)

data_dict = {'name':name_series, 'age':age_series}
dframe = pd.DataFrame(data_dict)   #create a pandas DataFrame from dictionary

dframe['age_plus_five'] = dframe['age'] + 5   #create a new column
dframe.pop('age_plus_five')
#dframe.pop('age')

salary = [1000,6000,4000,8000,10000]
salary_series = pd.Series(salary)
new_data_dict = {'name':name_series, 'age':age_series,'salary':salary_series}
new_dframe = pd.DataFrame(new_data_dict)
new_dframe['average_salary'] = new_dframe['age']*90

new_dframe.index = new_dframe['name']
print(new_dframe.loc['sam'])

  4. Save the file with the following method.

    1. Press Escape button
    2. Enter the command :wq

  5. Run the following command to run the job.

    /spark-submit --master yarn --deploy-mode cluster <filepath>/samplefile.py

    Screenshot showing how to run Spark submit job.

Monitor queries on an Apache Spark cluster in HDInsight on AKS

Spark History UI

  1. Click on the Spark History Server UI from the overview Tab.

    Screenshot showing Spark UI.

  2. Select the recent run from the UI using the same application ID.

    Screenshot showing how to run Spark UI.

  3. View the Directed Acyclic Graph cycles and the stages of the job in the Spark History server UI.

    Screenshot of DAG cycle.

Livy session UI

  1. To open the Livy session UI, type the following command into your browser https://<CLUSTERNAME>.<CLUSTERPOOLNAME>.<REGION>.projecthilo.net/p/livy/ui

    Screenshot of how to open Livy session UI.

  2. View the driver logs by clicking on the driver option under logs.

Yarn UI

  1. From the Overview Tab click on Yarn and, open the Yarn UI.

    Screenshot of how to select Yarn UI.

  2. You can track the job you recently ran by the same application ID.

  3. Click on the Application ID in Yarn to view detailed logs of the job.

    View Logs.

Reference