Skip to content

Work with environment variables and resource monitoring

An environment is the infrastructure used by the platform to store data and run computation. Once created, you can start working in the environments by creating pipelines and running them in the platform.

In addition, you can create and save environment variables that will allow you to parameterize certain variables in order to call them when running or deploying a pipeline.

Function name Method Return type Description
create_or_update_environment_variable CraftAiSdk.create_or_update_environment_variable (environment_variable_name, environment_variable_value) dict To create or update an environment variable available for all pipelines executions.
list_environment_variables CraftAiSdk.list_environment_variables() List of dict Get a list of all environments variables in the current environment.
delete_environment_variable CraftAiSdk.delete_environment_variable(environment_variable_name) dict Delete a specified environment variable.

Set up an environment variable

An environment variable is a value that can be passed to your pipeline code. It is used to store information that may be needed by the operating system or by applications that run on the platform (for example, by endpoints you deploy on the platform).

Tip

On this page, we will detail the usage of the SDK, but environment variables can be modified from a page in the web UI by going to: Environments > click on the three dots of the desired environment > Settings > Environment variables.

Create and update an environment variable

To create or update an environment variable available for all pipelines executions.

CraftAiSdk.create_or_update_environment_variable(environment_variable_name,
environment_variable_value)

Parameters

  • environment_variable_name (str) – Name of the environment variable to create.
  • environment_variable_value (str) – Value of the environment variable to create.

Returns

A dict object containing the ID of environment variable (with keys “id”)

Get the list of environment variables

Get the list of all environment variables in the current environment.

CraftAiSdk.list_environment_variables()

Parameter

No parameter

Returns

List of dicts of environment variables (with keys “name” and “value”)

Delete an environment variable

Delete a specified environment variable.

CraftAiSdk.delete_environment_variable(environment_variable_name)

Parameter

  • environment_variable_name (str) – Name of the environment variable to delete.

Returns

Dict (with keys “name” and “value”) of the deleted environment variable

Get resource metrics

Get resource metrics of the environment into dict format (by default) or in a .csv file. You can view these metrics in graph format using the Resource Metrics web page.

CraftAiSdk.get_resource_metrics(start_date, end_date, csv=False)

Parameter

  • start_date (datetime.datetime) - The beginning of the period.
  • end_date (datetime.datetime) - The end of the period.
  • csv (bool) - If True, it will return a csv file as bytes.

Returns

If csv is True, it will return bytes. Otherwise, dict with:

The resource metrics, with the following keys:

  • additional_data (dict): Additional data with the following keys:
    • total_disk (int): Total disk size in bytes.
    • total_ram (int): Total RAM size in bytes.
    • total_vram (int): Total VRAM size in bytes if there is a GPU.
    • cpu_usage (list of dict): The CPU usage in percent.
    • disk_usage (list of dict): The disk usage in percent.
    • ram_usage (list of dict): The RAM usage in percent.
    • vram_usage (list of dict): The VRAM usage in percent if there is a GPU.
    • gpu_usage (list of dict): The GPU usage in percent if there is a GPU.
    • network_input_usage (list of dict): The network input usage in bytes.
    • network_output_usage (list of dict): The network output usage in bytes.

Each element of the lists is a dict with the following keys:

  • metric (dict): Dictionary with the following key:
    • worker (str): The worker name.
  • values (list of list): The values of the metrics in the following format: [[timestamp, value], ...].

Example with dict object

from datetime import datetime, timedelta

# Get the current time
current_time = datetime.now()

# Calculate the beginning of the period (last 2 hours)
begin_date = current_time - timedelta(minutes=2)

# Set end_date to the current time
end_date = current_time

# Print the results
print("begin_date:", begin_date)
print("end_date:", end_date)

res = sdk.get_resource_metrics(begin_date, end_date)

print (res)

Example of return object:

{'metrics': {'cpu_usage': [{'metric': {'worker': 'Worker-58d65a25'},
    'values': [[1707132918930, 0.00464583333348214],
     [1707132948930, 0.004708333333450027],
     [1707132978930, 0.004763888888882371],
     [1707133008930, 0.0047222222219369045],
     [1707133038930, 0.004722222221995642]]},
   {'metric': {'worker': 'Worker-6e32691a'},
    'values': [[1707132918930, 0.007868055555652481],
     [1707132948930, 0.007638888888662494],
     [1707132978930, 0.00754861111112385],
     [1707133008930, 0.0076597222224098135],
     [1707133038930, 0.007930555555419681]]},

    ...

     [1707132948930, 40546.24444444444],
     [1707132978930, 48481.6111111111],
     [1707133008930, 40639.65555555555],
     [1707133038930, 48189.03333333333]]},
   {'metric': {'worker': 'All cumulate'},
    'values': [[1707132918930, 63759.26666666666],
     [1707132948930, 56443.922222222216],
     [1707132978930, 64879.22222222221],
     [1707133008930, 46783.055555555555],
     [1707133038930, 53954.188888888886]]}]},
 'additional_data': {'total_ram': 16616214528, 'total_disk': 21462233088}}

Example with CSV file

from datetime import datetime, timedelta

# Get the current time
current_time = datetime.now()

# Calculate the beginning of the period (last 2 hours)
begin_date = current_time - timedelta(minutes=2)

# Set end_date to the current time
end_date = current_time

# Print the results
print("begin_date:", begin_date)
print("end_date:", end_date)

res = sdk.get_resource_metrics(begin_date, end_date,csv=True)

result_csv = res.decode('utf-8')

# Save the result to a CSV file
csv_filename = "resource_metrics.csv"
with open(csv_filename, "w") as csv_file:
    csv_file.write(result_csv)

Example resource_metrics.csv file:

timestamp worker CPU RAM_percent RAM disk_percent disk network_input network_output
1707132918930 Worker-58d65a25 0.00464583333348214 0.0614234303655712 0.950531005859375 0.403168019866396 8.05862808227539 14646.7222222222 15505.5777777778
1707132948930 Worker-58d65a25 0.00470833333345003 0.0615619668532965 0.952674865722656 0.403168019866396 8.05862808227539 15029.0666666667 15897.6777777778
1707132978930 Worker-58d65a25 0.00476388888888237 0.0612654198875784 0.948085784912109 0.403217830899368 8.05962371826172 15577.8777777778 16397.6111111111
1707133008930 Worker-58d65a25 0.00472222222193691 0.061079554208317 0.945209503173829 0.40321668581815 8.05960083007813 5007 6143.4
1707133038930 Worker-58d65a25 0.00472222222199564 0.0612341135994269 0.947601318359374 0.40321668581815 8.05960083007813 4627.8 5765.15555555556
1707132918930 Worker-6e32691a 0.00786805555565248 0.0948272398231762 1.46745681762695 0.831449961000442 16.6192398071289 47515.6111111111 48253.6888888889
1707132948930 Worker-6e32691a 0.00763888888866249 0.0949657763109015 1.46960067749023 0.832236059256426 16.634952545166 39472.1555555555 40546.2444444444
1707132978930 Worker-6e32691a 0.00754861111112385 0.0951188566647759 1.47196960449219 0.832231097237816 16.6348533630371 47657.8444444444 48481.6111111111
1707133008930 Worker-6e32691a 0.00765972222240981 0.0950882898952422 1.47149658203125 0.832231097237816 16.6348533630371 39568.6777777778 40639.6555555556