• Install Python 3.8.13 in the machine learning service node.
  • Install pip3 and ensure it points to Python 3.8.13.
  • Install Python as the root user and make sure Python3 is pointing to python 3.8.13 (for example, by creating a soft link: ln -s /usr/local/bin/python3.8 /usr/bin/python3).
  • Make sure management port 8090 is available.

To install the PingIntelligence machine learning service:

  1. Go to the Ping Identity Product Downloads website.
  2. Under PingIntelligence for APIs, click View Now.
  3. Click Download under PingIntelligence for APIs.
  4. Select a deployment method.
  5. After downloading the file, copy the machine learning (ML) service file to the /opt directory or any other directory where you want to install the ML service.
  6. Change the working directory:
    • If you are installing as a root user, change the working directory to /opt.
    • If you want to install as a non-root user, choose any other location.
  7. At the command prompt, type the following command to untar the ML service build file: 
    tar –zxvf <filename>
    tar –zxvf pi-api-mlservice-5.2.tar.gz
  8. To verify that the ML service successfully installed, type the ls command at the command prompt. 
    /opt/pingidentity/$ ls pingidentity 
Ml_service

    This command lists the pingidentity directory and ml_service directory.

  9. Install the requirements: 
    pip3 install -r /opt/pingidentity/ml_service/lib/requirements.txt
  10. To update the properties.rc file:
    1. Add Kafka, ABS, and MongoDB details to the properties.rc file in config directory.
    2. Copy the MongoDB (mongo.crt) and Kafka (kafka.crt) certificates to the ml_service node and provide that path in the properties.rc file.
    3. Set export mongo_ssl to true. 
      export mongo_ssl="true"
    4. Set export kafka_producer_insync_replicas to 1. 
      export kafka_producer_insync_replicas=1