Python sample application - Amazon CodeGuru Profiler

Python sample application

In this tutorial, you’ll walk through the complete set up necessary to run Amazon CodeGuru Profiler within a sample application. You’ll then be able to view the profiling group's resulting runtime data.

To view the sample application, navigate to the CodeGuru Profiler Python demo application. CodeGuru Profiler runs inside the sample application and collects and reports profiling data about the application, which can be viewed from the AWS console.

This tutorial’s sample application does some basic image processing, with some CPU-heavy operations alongside some IO-heavy operations. It uses an Amazon Simple Storage Service bucket for cloud storage of images and an Amazon Simple Queue Service queue to order the names of images to be processed. It consists chiefly of two components which run in parallel, the task publisher and the image processor:

TaskPublisher checks the S3 bucket for available images, and submits the name of some of these images to the SQS queue.

ImageProcessor polls SQS for names of images to process. Processing an image involves downloading it from S3, applying some image transforms (e.g. converting to monochrome), and uploading the result back to S3.

Prerequisites

For this tutorial, you will need:

Step 1: Create a profiling group

Set up the required components by running the following AWS commands in your terminal.

  1. Configure the AWS CLI.

    When prompted, specify the AWS access key and AWS secret access key of the IAM user that you will use with CodeGuru Profiler.

    When prompted for the default Region name, specify the Region where you will create the pipeline, such as us-west-2.

    When prompted for the default output format, specify the .json.

    aws configure
  2. Create a profiling group in CodeGuru Profiler, named PythonDemoApplication.

    aws codeguruprofiler create-profiling-group --profiling-group-name PythonDemoApplication
  3. Create an Amazon SQS queue.

    aws sqs create-queue --queue-name DemoApplicationQueue
  4. Create an Amazon S3 bucket. Replace the YOUR-BUCKET with your desired bucket name.

    aws s3 mb s3://python-demo-application-test-bucket-YOUR-BUCKET
  5. Provision an IAM user. For information, see Create an IAM user or use one that is associated with your AWS account.

    Grant the IAM user access to AmazonSQSFullAccess, AmazonS3FullAccess, and AmazonCodeGuruProfilerFullAccess Amazon managed policies.

  6. Export the Amazon SQS URL, Amazon S3 bucket name, and the target Region.

    Remember to replace YOUR-AWS-REGION, YOUR-ACCOUNT-ID, and YOUR-BUCKET . YOUR-AWS-REGION should be the Region you specified in the AWS CLI configuration. YOUR-ACCOUNT-ID should be your AWS account ID. YOUR-BUCKET should be the bucket name you specified during Amazon S3 bucket creation.

    export DEMO_APP_SQS_URL=https://sqs.YOUR-AWS-REGION.amazonaws.com/YOUR-ACCOUNT-ID/DemoApplicationQueue export DEMO_APP_BUCKET_NAME=python-demo-application-test-bucket-YOUR-BUCKET export AWS_CODEGURU_TARGET_REGION=YOUR-AWS-REGION

Step 2: Set up the virtual environment

Create the virtual environment and install necessary resources by running the following commands in your terminal.

  1. Create and activate the virtual environment.

    python3 -m venv ./venv source venv/bin/activate
  2. Install boto3 and skimage. Installing scikit-image with Python 3.9 may cause failures. For more information, see Installing schikit-learn via pip.

    pip3 install boto3 scikit-image
  3. Install the CodeGuru Profiler agent.

    pip3 install codeguru_profiler_agent

Step 3: Run the application

Run the following command in your terminal to start the application. Then, verify that the application has begun profiling.

  1. Run the sample application with the CodeGuru Profiler Python Agent.

    Note that when running the demo application for the first time, it is expected to see error messages such as No messages exist in SQS queue at the moment, retry later. and Failed to list images in demo-application-test-bucket-1092734-YOUR-BUCKET under input-images/ printing to the terminal. Our demo application will handle image upload and SQS message publication after a few seconds.

    python3 -m codeguru_profiler_agent -p PythonDemoApplication aws_python_sample_application/main.py
  2. Verify that your profiling group is running.

    Navigate to the CodeGuru Profiler console.

    Choose Profiling groups.

    The PythonDemoApplication group should have status "Pending". You may need to wait a few minutes for this to update. After running the demo for approximately 15 to 20 minutes, the group should have status "Profiling", and you can view runtime data.

Step 4: Understanding the console

The CodeGuru Profiler console is where you can view the data that Profiler has gathered from running the application.

  1. Navigate to the CodeGuru Profiler console.

  2. Choose Profiling groups.

  3. Choose the PythonDemoApplication group.

The following sections are displayed, along with options for other visualizations or a full recommendations report.

  • Profiling group status displays the status of the profiling group and metrics from data collected in the past 12 hours.

  • CPU summary displays the amount of system CPU capacity that the application consumes. You can choose Visualize CPU to view the flame graph.

    
              Image: CPU visualization of demo application.
  • Latency summary displays the amount of time the application’s threads spend in the Blocked, Waiting, and Timed Waiting thread states.

  • Heap usage displays how much of your application's maximum heap capacity is consumed by the application.

  • Anomalies display any deviations from trends that CodeGuru Profiler detects.

  • Recommendations display suggestions to optimize the application.

Cleanup

Upon completion of this tutorial, clean up the resources created.

Delete the profiling groups by following the procedure in Deleting a profiling group.