FEA-Bench / testbed /deepset-ai__haystack /releasenotes /notes /datadog-tracer-b084cf64fcc575c6.yaml
| features: | |
| - | | |
| Added out-of-the-box support for the Datadog Tracer. This allows you to instrument pipeline and component | |
| runs using Datadog and send traces to your preferred backend. | |
| To use the Datadog Tracer you need to have the `ddtrace` package installed in your environment. | |
| To instruct Haystack to use the Datadog tracer, you have multiple options: | |
| * Run your Haystack application using the `ddtrace` command line tool as described in the | |
| the [ddtrace documentation](https://ddtrace.readthedocs.io/en/stable/installation_quickstart.html#tracing). | |
| This behavior can be disabled by setting the `HAYSTACK_AUTO_TRACE_ENABLED_ENV_VAR` environment variable to `false`. | |
| * Configure the tracer manually in your code using the `ddtrace` package: | |
| ```python | |
| from haystack.tracing import DatadogTracer | |
| import haystack.tracing | |
| import ddtrace | |
| tracer = ddtrace.tracer | |
| tracing.enable_tracing(DatadogTracer(tracer)) | |
| ``` | |