Note: this sample is based on the source code exploration and is not officially documented as a way to setup OpenTelemetry for Prompt Flow applications.
- Install dependencies:
pip install -r requirements.txt- Create an Azure OpenAI connection:
pf connection create --file ./connections/azure_openai.yaml --set api_key=<your API key> api_base=<API base URL>-
Create an
.envfile based on the.env.examplefile and fill with correct values for your environment -
Run the flow:
pf flow test --flow .\chat-with-wikipedia\Instrumentation is setup in the instrumentation.py file and is based on Azure Monitor Opentelemetry Distro client
library for
Python.
By importing the instrumentation module in PF nodes, we make it execute the configure_tracing_and_logging() top level
function.
This function configures the OpenTelemetry SDK with the Azure Monitor exporter and sets up the tracing and logging that is subsequently used by the Prompt Flow for instrumentation of nodes.
