An example Azure Function App that includes an HTTP trigger Azure Function instrumented using Open Telemetry.
When the URL for the function is requested, distributed tracing (DT) data is extracted from the request headers and applied to a newly-created span to represent the function.
When debugging locally, the URL to invoke will be output to the terminal. For example: [GET,POST] http://localhost:7071/api/HttpTriggerExample. This can then be invoked via command-line, web browser, etc. Please keep in mind the VS Code Azure debugging tools don't always process things correctly. Sometimes query params don't get processed, for example.
Once you've created an Azure Function App in Azure and deployed, you can grab the Azure URL via the VS Code command pallet: Command + Shift + P -> Azure Functions: Copy Function URL. See the Deployment section below.
This example currently only exports trace data and does not have metric examples.
Traces are buffered using a BatchSpanProcessor. The batch span processor will export when either the timeout is hit, the buffer size limit is hit, or the function completes execution via a call to force flush.
A minimal amount of useful HTTP specific data has been captured for this example. You may wish to capture additional details and add according to the HTTP Semantic Conventions.
See also the FAAS Semantic conventions, which are currently experimental and subject to change:
- https://github.com/open-telemetry/semantic-conventions/blob/main/docs/resource/faas.md
- https://github.com/open-telemetry/semantic-conventions/blob/main/docs/faas/faas-spans.md
The following settings are recommended to be set in your Azure Function App configuration and local settings (local.settings.json) or ENV vars. The first time you debug an example, VS Code will automatically create local.settings.json for you. This file is not committed to source control as it may contain secrets.
-
OTEL_EXPORTER_OTLP_ENDPOINT: Must be set to a valid endpoint such as:https://otlp.nr-data.net:4317. See options in the Open Telemetry Setup page. -
API_KEY[required]: Must be set to allow data to be sent to new relic. Requires a new relic ingest license key: https://one.newrelic.com/launcher/api-keys-ui.launcher.
Instead of manually plumbing values configuration as in the example, the following environment variables may be set to configure export to New Relic and adhere to New Relic's data ingest limits:
OTEL_EXPORTER_OTLP_HEADERS="api-key=<your_license_key>"- Replace
<your_license_key>with your Account License Key.
- Replace
OTEL_ATTRIBUTE_VALUE_LENGTH_LIMIT=4094
npm install from the http-trigger-app folder.
You'll likely need to setup/point-to an Azure Storage account your first time debugging. Multiple files in .vscode have been committed to source control to aid in debugging the examples using VS Code, which includes the debug launch settings.
Use npm run dev to run locally, without a debugger, with multiple azure functions apps running at the same time. otherwise, they will conflict on port when launching.
Creating a new Azure Function App can be done through the website or via the Azure extension. Below describes creating through the Azure extension in VS Code.
[Resources -> Function App -> right-click] OR [command pallet (Command + Shift + P)] -> Azure Functions: Create Function App in Azure....
Enter a name, a Node.js runtime and a region. This current version of the example has been tested with Node 16 in Azure and Node 18 locally.
Add required application settings as well as any desired optional settings. These will become ENV variables available to the the function.
An easy way to get started is through the Azure plugin. Select the function app -> Application Settings -> right-click -> Upload local settings OR use the command pallet (Command + Shift + P) -> Azure Functions: Upload Local Settings... And then modify any different settings you need in Azure. Modification can also be done through the Azure extension.
Now in your primary application folder, deploy the application to the Azure Function App.
[right-click] OR [command pallet (Command + Shift + P)] -> Azure Functions: Deploy to Function App... and choose the Function App you created earlier.
Once deployed, to execute from another function within Azure, you'll need to get the proper URL.
This can be done inside VS Code by leveraging the command pallet. Command + Shift + P -> Azure Functions: Copy Function URL. Choose the deployed Azure Function App and then the specific function. Once selected, the URL will be on your clip-board. Paste into the appropriate location for usage, such as the EXTERNAL_URL setting of the queue trigger example.
This URL will likely take the form https://<azure function app>.azurewebsites.net/api/<functionname>?code=<some code>.
You can also get the URL through the Azure UI. Navigate to your Azure Function App -> Functions (left nav) -> HttpTriggerExample -> Get Function URL (tool bar) -> copy the default function key.