Skip to content

lumigo-io/python_tracer

Repository files navigation

CircleCI Version codecov

This is lumigo/python_tracer, Lumigo's Python agent for distributed tracing and performance monitoring.

Supported Python Runtimes: 3.6, 3.7, 3.8, 3.9, 3.10, 3.11 and 3.12

Usage

The package allows you to pursue automated metric gathering through Lambda Layers, automated metric gathering and instrumentation through the Serverless framework, or manual metric creation and implementation.

With Lambda layers

  • When configuring your Lambda functions, include the appropriate Lambda Layer ARN from these tables

Note - Lambda Layers are an optional feature. If you decide to use this capability, the list of Lambda layers available is available here..

Learn more in our documentation on auto-instrumentation.

With Serverless framework

Manually

To manually configure Lumigo in your Lambda functions:

  • Install the package:
pip install lumigo_tracer
  • Import the package in your Lambda code:
`from lumigo_tracer import lumigo_tracer`
  • Next, wrap your handler in Lumigo's trace function (note: replace YOUR-TOKEN-HERE with your Lumigo API token):
@lumigo_tracer(token='YOUR-TOKEN-HERE')
def my_lambda(event, context):
    print('I can finally troubleshoot!')
  • Your function is now fully instrumented

Configuration

@lumigo/python_tracer offers several different configuration options. Pass these to the Lambda function as environment variables:

  • LUMIGO_DEBUG=TRUE - Enables debug logging
  • LUMIGO_SECRET_MASKING_REGEX=["regex1", "regex2"] - Prevents Lumigo from sending keys that match the supplied regular expressions. All regular expressions are case-insensitive. By default, Lumigo applies the following regular expressions: [".*pass.*", ".*key.*", ".*secret.*", ".*credential.*", ".*passphrase.*"].
    • We support more granular masking using the following parameters. If not given, the above configuration is the fallback: LUMIGO_SECRET_MASKING_REGEX_HTTP_REQUEST_BODIES, LUMIGO_SECRET_MASKING_REGEX_HTTP_REQUEST_HEADERS, LUMIGO_SECRET_MASKING_REGEX_HTTP_RESPONSE_BODIES, LUMIGO_SECRET_MASKING_REGEX_HTTP_RESPONSE_HEADERS, LUMIGO_SECRET_MASKING_REGEX_HTTP_QUERY_PARAMS, LUMIGO_SECRET_MASKING_REGEX_ENVIRONMENT.
  • LUMIGO_DOMAINS_SCRUBBER=[".*secret.*"] - Prevents Lumigo from collecting both request and response details from a list of domains. This accepts a comma-separated list of regular expressions that is JSON-formatted. By default, the tracer uses ["secretsmanager\..*\.amazonaws\.com", "ssm\..*\.amazonaws\.com", "kms\..*\.amazonaws\.com"]. Note - These defaults are overridden when you define a different list of regular expressions.
  • LUMIGO_PROPAGATE_W3C=TRUE - Add W3C TraceContext headers to outgoing HTTP requests. This enables uninterrupted transactions with applications traced with OpenTelemetry.
  • LUMIGO_SWITCH_OFF=TRUE - In the event a critical issue arises, this turns off all actions that Lumigo takes in response to your code. This happens without a deployment, and is picked up on the next function run once the environment variable is present.

Step Functions

If your function is part of a set of step functions, you can add the flag step_function: true to the Lumigo tracer import. Alternatively, you can configure the step function using an environment variable LUMIGO_STEP_FUNCTION=True. When this is active, Lumigo tracks all states in the step function in a single transaction, easing debugging and observability.

@lumigo_tracer(token='XXX', step_function=True)
def my_lambda(event, context):
    print('Step function visibility!')

Note: the tracer adds the key "_lumigo" to the return value of the function.

If you override the "Parameters" configuration, add "_lumigo.$": "$._lumigo" to ensure this value is still present.

Below is an example configuration for a Lambda function that is part of a step function that has overridden its parameters:

"States": {
    "state1": {
      "Type": "Task",
      "Resource": "arn:aws:lambda:us-west-2:ACCOUNT:function:FUNCTION_NAME",
      "Parameters": {
          "Changed": "parameters",
          "_lumigo.$": "$._lumigo"
        },
      "Next": "state2"
    },
    "state2": {
      "Type": "pass",
      "End": true
    }
}

Logging Programmatic Errors

Lumigo provides the report_error function, which you can use to publish error logs that are visible to the entire platform. To log programmatic errors:

  • Import the report_error function with the following code: from lumigo_tracer import report_error
  • Use the report_error function with the message you wish to send: report_error("your-message-here")

Adding Execution Tags

You can add execution tags to a function with dynamic values using the parameter add_execution_tag.

These tags will be searchable from within the Lumigo platform.

Limitations

  • Up to 50 execution tags
  • Each tag key length can have 50 characters at most.
  • Each tag value length can have 70 characters at most.

Contributing

Contributions to this project are welcome from all! Below are a couple pointers on how to prepare your machine, as well as some information on testing.

Preparing your machine

Getting your machine ready to develop against the package is a straightforward process:

  1. Clone this repository, and open a CLI in the cloned directory
  2. Create a virtual environment for the project virtualenv venv -p python3
  3. Activate the virtualenv: . venv/bin/activate
  4. Install dependencies: pip install -r requirements.txt
  5. Run the setup script: python setup.py develop.
  6. Run pre-commit install in your repository to install pre-commit hooks

Note: If you are using pycharm, ensure that you set it to use the virtualenv virtual environment manager. This is available in the menu under PyCharm -> Preferences -> Project -> Interpreter

Running the test suite

We've provided an easy way to run the unit test suite:

  • To run all unit tests, simply run py.test in the root folder.
  • To deploy services for component tests, run sls deploy from the root test directory. This only needs to take place when the resources change.
  • To run component tests, add the --all flag: py.test --all