faropt package

Module contents

class faropt.FarOpt(framework='ortools', stackname='faropt')

Bases: object

FarOpt class used to initialize, configure and submit jobs to the back end :param framework: Currently only ortools, TO DO is to extend to other frameworks. Note that other frameworks like pyomo, DEAP, inspyred and pulp are supported :type framework: string, optional :param stackname: Points to the backend CDK stack that needs to be launched separately. Default name is faropt, but you many need to pass in another name while testing :type stackname: string, optional

add_recipe(recipe_name, maintainer='Faropt SDK user')

Adds a recipe referencing the job that you submitted (see self object params).

Parameters

  • recipe_name (string) – Friendly name for your recipe

  • maintainer – Recipe author/maintainer

configure(source_dir)

Zips up a local folder containing your main.py code, and any other subfolders/files required to run your project. Make note of the output structure printed to see if all files that you need are printed.ArithmeticError

Parameters

source_dir (string) – path to your source, such as ‘./home/src/’

get_metric_data(metric_name)

Returns raw metric data that was submitted from the backend. To use this, do from utils import * in your main.py, and then use log_metric like this, for e.g: log_metric(‘total_distance’,total_distance)

Returns

response from cloudwatch

Return type

json string

get_recipe_id_from_description(description)

Returns UUID of a recipe based on friendly description/ recipe name

Parameters

description (string) – friendly description/ recipe name

Returns

First UUID that matches the description of the recipe

Return type

uuid4()

list_jobs(limit=10, verbose=True)

Returns list of jobs submitted

Parameters

  • limit (int, optional) – Number of jobs to return, Defaults to 10

  • verbose (bool, optional) – Verbose print of the job table, Defaults to True

Returns

job table scan (raw) results

Return type

boto3 response

list_recipes(limit=10, verbose=True)

Returns list of recipes registered

Parameters

  • limit (int, optional) – Number of recipes to return, Defaults to 10

  • verbose (bool, optional) – Verbose print of the recipe table, Defaults to True

Returns

Recipe table scan (raw) results

Return type

boto3 response

logs()

Prints logs of a submitted job.

primary_status()

Returns the last status of the submitted job; Can be PROVISIONING > PENDING > RUNNING > STOPPED > JOB COMPLETED

Returns

primary staus

Return type

string

printlogs(response)
run_recipe(recipe_id, micro=False)

Runs already registered recipe

Parameters

  • recipe_id (bool) – UUID of recipe

  • micro – Submit as a micro job

run_s3_job(bucket, key, micro=False)

Runs job based on a source file in bucket/key. For example, place a source.zip in s3://bucket/key/source.zip and submit a job

Parameters

  • bucket (string) – Bucket name

  • key (string) – path/key on S3 that looks like path/to/s3/key/source.zip inside the bucket

  • micro (bool) – Submit a micro job.

stage()

Uploads the source.zip but does not submit to fargate. Useful when you want to run later

status()

Returns the full status of the submitted job; used in primary_status, which should be enough for most use cases

stop()

Stops a submitted task

stream_logs(start_time=0, skip=0)

Internal, use self.logs() instead of streaming

submit(micro=False)

Runs job defined in object params. Creates a new job ID to track and sets submitted to True. Check self.jobname to reference the job that was submitted. View self.logs() once the job has completed

Parameters

micro (bool) – Submit a micro job. By submitting a micro job, you are restricted to using ortools, pyomo and deap libraries for jobs that last up to 5 minutes

wait()

Polls for the primary status of the container task that runs this job. You should see PROVISIONING > PENDING > RUNNING > STOPPED > JOB COMPLETED

Returns

primary status of the job that was submitted

Return type

list