components module

components.autoscale_sensors(iam_role, s3_bucket_name, ansys_license_server='', virtual_sensor_img='', cpu=2, memory=2048, gpu=0, s3_IoT_key='DG-NG-sensor-data/')

Automatically create or destroy virtual sensors in an AWS Batch environment. This function can be used with or without an ANSYS license server. :rtype: None

  1. A connection to the Batch compute environment is established.

  2. The general job definition for a virtual sensor is defined

  3. Existing running virtual sensors (i.e. existing Batch jobs) are queried

  4. New sensor is checked in an S3 Bucket

    Objects are assumed to be named: <someName>_#.<csv> All of the object # are determined and assumed that a virtual sensor will be associated with each #. If there is no virtual sensor from step 3 associated with #, then a virtual sensor will be created. If there is a virtual sensor and no associated #, then the sensor will be destroyed.

  5. Kill all sensors no longer associated with data.

  6. Create all new sensors that do not exist, but data is now available.

Parameters:

  • iam_role (str) – IAM rol to be assumed by virtual sensors.

  • s3_bucket_name (str) – The s3 bucket to look for data. #TODO: will add connections to all other TwinModules data sources. For meta data determination.

  • ansys_license_server (str, optional) – Use this license server ip address if applicable. The default is “”.

  • virtual_sensor_img (str, optional) – The container image to use (URL). The default is “”.

  • cpu (int, optional) – Required cpus needed for virtual sensor. The default is 2.

  • memory (int, optional) – Required RAM in MB needed for virtual sensor. The default is 2048.

  • gpu (int, optional) – Number of gpus needed by virtual sensor. Note the AWS Batch compute environment must be able to support gpus. The default is 0.

  • s3_IoT_key (str, optional) – The s3 key/folder to use for new data. The default is “DG-NG-sensor-data/”.

Return type:

None