Skip to content

Error tracking

The orchestrator-core supports Sentry for error tracking and performance monitoring. Sentry is an application monitoring platform that helps developers identify, debug, and resolve issues in their applications by providing real-time error tracking, performance monitoring, and distributed tracing capabilities.

In order to initialize Sentry (assuming you have already set up a Sentry project), perform the following steps:

1. Update your own Settings class

Add the following attributes to the Settings object in your own orchestrator's settings.py:

TRACING_ENABLED: bool = False
SENTRY_DSN: str = ""
TRACE_SAMPLE_RATE: float = 0.1

# settings.py
from pydantic_settings import BaseSettings

class MySettings(BaseSettings):
    TRACING_ENABLED: bool = False
    SENTRY_DSN: str = ""
    TRACE_SAMPLE_RATE: float = 0.1

my_settings = MySettings()

2. Set environment variables 

TRACING_ENABLED=True
SENTRY_DSN = "your_sentry_dsn" # should be obtained from Sentry
TRACE_SAMPLE_RATE = 0.1 # should be a float between 0 and 1
Setting TRACING_ENABLED to True will enable tracing for the application, allowing you to monitor performance and errors more effectively.

Note

  • SENTRY_DSN: The Data Source Name (DSN) is a unique URL provided by Sentry. It connects your application to your Sentry project so errors and performance data are sent to the correct place.
  • TRACE_SAMPLE_RATE: A float between 0 and 1 that controls what percentage of transactions are sent to Sentry for performance monitoring (e.g., 0.5 means 50% of traces are sampled). See Sentry documentation for more details on sampling.

3. Update main.py file

Add the following code to the main.py file of the orchestrator-core application: 

from orchestrator import OrchestratorCore
from orchestrator.cli.main import app as core_cli
from orchestrator.settings import AppSettings
from my_orchestrator.settings import my_settings

app = OrchestratorCore(base_settings=AppSettings())

if app.base_settings.TRACING_ENABLED and app.base_settings.ENVIRONMENT != "local":
    from orchestrator.app import sentry_integrations
    from sentry_sdk.integrations.httpx import HttpxIntegration

    sentry_integrations.append(HttpxIntegration())

    app.add_sentry(
        my_settings.SENTRY_DSN,
        my_settings.TRACE_SAMPLE_RATE,
        app.base_settings.SERVICE_NAME,
        app.base_settings.ENVIRONMENT,
    )

if __name__ == "__main__":
    core_cli()

Note

  • It's recommended to use separate Sentry projects for different environments (production, staging, etc.) to avoid cluttering production error tracking with development errors. Many teams skip Sentry integration in local development environments.
  • Make sure to adjust the TRACE_SAMPLE_RATE according to your needs and test the integration in a development environment before deploying it to production.

4. Test the Sentry integration

You can test the Sentry integration by triggering an error in your application. For example, you can create a route that raises an exception when the endpoint is triggered:

from fastapi import FastAPI

app = FastAPI()

@app.get("/debug-sentry")
def debug_sentry():
    division_by_zero = 1 / 0

Warning

Make sure your environment variable ENVIRONMENT is NOT set to local when testing the Sentry integration.

See also