renops-scheduler 1.1.3

Scheduler shifts given scripts to time when most renewables available

renops-scheduler

renops-scheduler is a Python package that allows you to schedule and execute scripts at the time the most renewable energy is available.

This reduces the carbon footprint and stabilises the grid of our operations. In cases where the electricity price is variable, the scheduler also reduces utility costs. While its main focus are energy-intensive processing tasks such as big data analytics or AI model training, it may be used to shift any energy-intensive load such as HVAC.

Requirements

Before you begin, ensure you have met the following requirements:

• Python: Version 3.8 or higher.

To check your Python version, run:

$ python3 --version

Installation

To install renops-scheduler, run the following command:

$ pip install renops-scheduler

Usage

Once you have installed renops-scheduler, you can use it to schedule and execute Python scripts in command line interface (CLI).

To use the program, follow these steps:

1. Open a terminal or command prompt.

2. Create a new file named test.py containing:

   $ echo 'print("hello world!")' > test.py
   

3. Export RENOPSAPI_KEY env:

   $ export RENOPSAPI_KEY="TANGO_DEMO_KEY" 
   

   NOTE: This is demo key with limited number of request. Contanct us to obtain personal access token.

4. Run the following command to execute the script with a deadline of 24 hours:

   $ renops-scheduler test.py -la -r 6 -d 24 -v --optimise renewable
   

   This will execute the test.py in an optimal window within the given deadline.

   • -la sets automatic location detection (uses machines public IP!),
   • -r 6 sets runtime (estimated by user),
   • -d 24 sets the deadline to 24 hours.

5. Scheduler can also find interval with minimal energy price:

   $ renops-scheduler test.py -la -r 6 -d 24 -v --optimise price
   

   This is achieved by adding --optimise price flag.

6. Running scheduler without automatic location detection:

   $ renops-scheduler test.py -l "Berlin,Germany" -r 6 -d 24 -v
   

   In cases where a user does not want to expose its IP, due to privacy concerns, we can manually specify a rough location in a text description.

7. Scheduler can be used to find inteval with minimal carbon emissions:

   $ renops-scheduler test.py -la -r 6 -d 24 -v --optimise emissions
   

   This is achieved by adding --optimise emissions flag.

Import Example

from renops.scheduler import Scheduler

# Define a function with an argument that scheduler will execute
def test_run(a, text: str):
    print("Hello World!")
    print("Passed keyword argument:", text)
    print("Passed argument:", a)

# Intialise the scheduler
s = Scheduler(runtime=1,
              deadline=1,
              location="Kranj",
              verbose=True,
              optimise_type="price",
              action=test_run,
              argument=([42]),
              kwargs={"text": "Scheduler Test!"})

Geographical Shifting

Scheduler allows you to define a set of available endpoints with their associated commands for energy-intensive tasks. The tool intelligently analyzes these locations and selects the optimal one based on your specified metrics (such as renewable energy potential, price, etc.).

CLI

Populate locations.json with following content:

{
    "hpc1": {
        "location": "Berlin, Germany",
        "cmd": "ssh user@hpc1 python3 test.py"
    },
    "hpc2": {
        "location": "Madrid, Spain",
        "cmd": "ssh user@hpc2 python3 test.py"
    },
    "hpc3": {
        "location": "Copenhagen, Denmark",
        "cmd": "ssh user@hpc3 python3 test.py"
    }
}

Run scheduler by passing path to geo-shift.json together with --geo-shift flag

$ renops-scheduler locations.json --geo-shift --optimise-price --verbose

Scheduler will find optimal location to execute the script based on given metric.

Python Script

from renops.geoshifter import GeoShift

# Define locations and commands for each endpoint
locations = {
    "hpc1": {
        "location": "Berlin, Germany",
        "cmd": "ssh user@hpc1 python3 test.py"
    },
    "hpc2": {
        "location": "Madrid, Spain",
        "cmd": "ssh user@hpc2 python3 test.py"
    },
    "hpc3": {
        "location": "Copenhagen, Denmark",
        "cmd": "ssh user@hpc3 python3 test.py"
    }
}

# Intialise the shifter
gs = GeoShift(
    locations=locations,
    optimise_type="price",
    verbose=True
)

# Run geoshifter
gs.shift()

Arguments

The program accepts several command-line arguments to customize the execution. Here's an overview of the available options:

usage: renops-scheduler [-h] -l LOCATION [-gs] [-o {renewable,price,emissions}] [-v] [-r RUNTIME] [-d DEADLINE] script_path

positional arguments:
  script_path           Path to the script to be executed or JSON file in case of geo shifting.

options:
  -h, --help            show this help message and exit
  -l LOCATION, --location LOCATION
                        Location can be specified in two ways:
                        
                        1. Pass a specific location as a string, e.g., "Berlin, Germany".
                        
                        2. Use automatic location detection based on IP address.
                         By using this tag, you agree that your IP can be used to detect your location.
                        You can use any of the following values for this purpose:
                           -l a (-la)
                           -l auto
                           -l automatic
 
  -o {renewable,price,emissions}, --optimise {renewable,price,emissions}
                        Choose an optimisation type:
                         - 'renewable' (renewable potential - renewable energy availability on a scale from 0 to 1)
                         - 'price' (day-ahead energy price)
                         - 'emissions' (Carbon emissions in gCO2eq/kWh)

  -gs, --geo-shift      JSON on given path should be formated as:
                        {
                          "hpc1": {
                            "location": "Berlin, Germany",
                            "cmd": "ssh user@hpc1 python3 train.py"
                          },
                          "hpc2": {
                            "location": "Madrid, Spain",
                            "cmd": "ssh user@hpc2 python3 train.py"
                          },
                          "hpc3": {
                            "location": "Copenhagen, Denmark",
                            "cmd": "ssh user@hpc3 python3 train.py"
                          }
                        }
 
  -v, --verbose         Verbose mode.
  -r RUNTIME, --runtime RUNTIME
                        Runtime in hours. (Not for geo shift mode)
  -d DEADLINE, --deadline DEADLINE
                        Deadline in hours, by when should script finish running (Not for geo shift mode)

Privacy

The script does not pose security or privacy concerns, as it runs locally. The communication between our forecasting API is encrypted and includes an approximate location of the requested forecast. Automatic localization through IP is mandatory and must be manually set.

Licences

This project is licensed under the Apache License - see the LICENSE file for details.

Attributions

• Open meteo (https://open-meteo.com/en/license)
• NASA POWER API (https://power.larc.nasa.gov/docs/services/api/)
• ENTSO-e (https://transparency.entsoe.eu/content/static_content/download?path=/Static%20content/terms%20and%20conditions/231018_List_of_Data_available_for_reuse.pdf)
• OpenStreetMap (https://osmfoundation.org/wiki/Licence)

Notes

• renops-scheduler is currently in beta version and may contain bugs or limitations.
• Send possible suggestions, bugs and improvements to [e-mail]