Datadog dd-trace-py Changelog File — Datadog

Changelog

Changelogs for versions not listed here can be found at https://github.com/DataDog/dd-trace-py/releases

3.8.0

New Features

LLM Observability: add processor capability to process span inputs and outputs. See usage documentation [here](https://docs.datadoghq.com/llm_observability/setup/sdk/python/#span-processing).
CI Visibility: This introduces the ability to gzip the payload when using the evp proxy setup, incurring in less network bandwith consumption.
Error Tracking: Introduces automatic reporting of handled exceptions. Enabling the feature will report handled exceptions to Error Tracking from the user code, the third party packages code, some specified modules or everything based on configuration. This feature can be controlled using two environment variables: - DD_ERROR_TRACKING_HANDLED_ERRORS=`allthird_party` - DD_ERROR_TRACKING_HANDLED_ERRORS_INCLUDE=`module1, module2, module3.submodule`
Code Security: IAST support for langchain v0.1.0 and above.
openai: This introduces tracing support for the OpenAI Responses endpoint.

Bug Fixes

tracing: Fixes an issue where truncation of span attributes longer than 25000 characters would not consistently count the size of UTF-8 multibyte characters, leading to a unicode string is too large error.
CI Visibility: This fix resolves an issue where the DD_CIVISIBILITY_ITR_ENABLED was not honored properly.
tracing: Fixes a bug in distributed tracing where pickling ddtrace.trace.Context fails in coroutines. This regression was introduced in v3.7.0.
CI Visibility: This fix resolves an issue where pytest-xdist would not exit with the proper status code if ATR was enabled.
CI Visibility: This fix resolves an issue where ddtrace pytest plugin used with xdist would report test suites as failing even when all tests pass.
profiling: fixed an issue in the SynchronizedSamplePool where pool could be null when calling into ddog_ArrayQueue_ functions, leading to segfaults in the uWSGI shutdown
Code Security: IAST: Avoid excessive filtering of stacktrace locations when finding vulnerabilities. After this change, vulnerabilities that were previously discarded will now be reported. In particular, if they were found within code in site-packages or outside of the working directory.
LLM Observability: Resolves an issue where spans and evaluation metrics were not being sent via Unix sockets.
dynamic instrumentation: prevent an exception when trying to remove a probe that did not resolve to a valid source code location.
kafka: This fix resolves an issue where message headers were sent to Kafka brokers that do not support them. Message headers are turned off when the Kafka server responds with UNKNOWN_SERVER_ERROR (-1).
code origin for spans: fixes a performance issue with exit spans.
profiling: improve performance of the memory profiler for large heaps. The memory profiler previously did a linear search of tracked allocations for every free, which scaled very poorly with large heaps. Switch to a fast hash map.

3.7.2

Bug Fixes

CI Visibility: This fix resolves an issue where the DD_CIVISIBILITY_ITR_ENABLED was not honored properly.
tracing: Fixes an issue where truncation of span attributes longer than 25000 characters would not consistently count the size of UTF-8 multibyte characters, leading to a unicode string is too large error.

3.7.1

Bug Fixes

CI Visibility
- Resolves an issue where pytest-xdist would not exit with the proper status code if ATR was enabled.
- Resolves an issue where ddtrace pytest plugin used with xdist would report test suites as failing even when all tests pass.

2.21.8

Bug Fixes

Code Security:
- Avoid excessive filtering of stacktrace locations when finding vulnerabilities. After this change, vulnerabilities that were previously discarded will now be reported. In particular, if they were found within code in site-packages or outside of the working directory.
- Fixes a bug where invalid f-strings didn’t raise the expected "Unknown format code" error when IAST was enabled.
Profiling:
- Improve performance of the memory profiler for large heaps. The memory profiler previously did a linear search of tracked allocations for every free, which scaled very poorly with large heaps. Switch to a fast hash map.
Other:
- Fix a potential circular import with the psycopg2 contrib.
- Code origin for spans: fixes a performance issue with exit spans.

2.21.6

Bug Fixes

Code Security
- Fixes an issue with PosixPath handling in path operations that could cause errors during taint tracking. This fix improves stability and slightly reduces import times.

3.2.3

Bug Fixes

Code Security
- Fixes an issue with PosixPath handling in path operations that could cause errors during taint tracking. This fix improves stability and slightly reduces import times.
Lib-injection
- Avoids zombie process from telemetry sender on startup.
LLM Observability
- Fixes an issue where LLMObs could not be enabled in a forked process when setting agentless_enabled=True or DD_LLMOBS_AGENTLESS_ENABLED=true.
Tracing
- internal: Fixes an issue where trimming a traceback to attach it to the span could result in the loss of the most recent frames.

3.3.3

Bug Fixes

Code Security -Fixes an issue with PosixPath handling in path operations that could cause errors during taint tracking. This fix improves stability and slightly reduces import times.

3.2.1

Bug Fixes

Library Injection
- Fix for release script causing lib injection OCI images to not get published.

3.3.1

Bug Fixes

ASM
- Fixes a NotImplementedError that occurred when trying to deepcopy wrapped builtin functions (like open) while ASM or IAST were enabled. The error was caused by the wrapper not implementing the __deepcopy__ method.
CI Visibility
- Resolves an issue where JUnit XML output would not count tests retried by Early Flake Detection, Auto Test Retries, and Attempt-to-Fix.
Lib-Injection
- Avoids zombie process from telemetry sender on startup.
LLM Observability
- Resolves an issue where large spans traced within a short time interval were dropped despite being under the 1 MB limit.

2.21.5

Bug Fixes

ASM
- Fixes a NotImplementedError that occurred when trying to deepcopy wrapped builtin functions (likeopen) while ASM or IAST were enabled. The error was caused by the wrapper not implementing the __deepcopy__ method.
LLM Observability
- Fixes an issue where LLMObs could not be enabled in a forked process when setting agentless_enabled=True or DD_LLMOBS_AGENTLESS_ENABLED=true.
Profiling
- Resolves an issue where the Lock profiler would throw an AttributeError: '_ProfiledThreadingLock' object has no attribute '_self_acquired_at'.
Tracing
- pylibmc: Fixes an issue where using Client(server=[url]) would throw the error __init__() got multiple values for argument 'servers'

Other Changes

library: Ensures that the SSI is not used for uWSGI applications. For enablement instructions, refer to the following our advanced_usage docs.

3.3.2

Bug Fixes

Lib-Injection
- Avoids zombie process from telemetry sender on startup.
LLM Observability
- Resolves an issue where large spans traced within a short time interval were dropped despite being under the 1 MB limit.
- Resolves an issue with anthropic LLM spans where multiple system prompts caused missing input messages.
- Fixes an issue where LLMObs could not be enabled in a forked process when setting agentless_enabled=True or DD_LLMOBS_AGENTLESS_ENABLED=true.

Other Changes

library: Ensures that the SSI is not used for uWSGI applications. For enablement instructions, refer to the following our advanced_usage docs .

3.4.1

Bug Fixes

Code Security
- Fixes an issue with PosixPath handling in path operations that could cause errors during taint tracking. This fix improves stability and slightly reduces import times.

3.4.0

Upgrade Notes

Profiling
- Upgrades echion which includes performance improvements.

Deprecation Notes

Tracing
- patch_all is deprecated. As an alternative to patch_all, you can use import ddtrace.auto along with DD_PATCH_MODULES if specific module patching is necessary.

New Features

Dynamic Instrumentation
- Adds support for the Decimal type from the Python standard library.
LLM Observability
- Introduces a integration tag to LLM Observability spans that are generated by an integration.
- agentless_enabled and DD_LLMOBS_AGENTLESS_ENABLED now defaults to None, and will be determined automatically based on the presence of a compatible Datadog agent. agentless_enabled and DD_LLMOBS_AGENTLESS_ENABLED can still be set explicitly to True or False</span>, but is now optional for agentless users.
Profiling
- Stack v2 supports CPython 3.13.
- Adds support for native exporter on Windows.
- Enables profiling native modules for macOS x86_64.
Tracing
- Adds configuration for encoding span events as a top-level field in v0.4 payloads and introduces environment variable DD_TRACE_NATIVE_SPAN_EVENTS (disabled by default). This requires agent version 7.63.0 or later.
- Extracts the referrer hostname from HTTP requests and stored it as http.referrer_hostname tag.

Bug Fixes

Fixes an issue with gevent support and the typing module on CPython >= 3.12.
ASM
- Fixes a NotImplementedError that occurred when trying to deepcopy wrapped builtin functions (like open) while ASM or IAST were enabled. The error was caused by the wrapper not implementing the __deepcopy__ method.
- SCA: Resolves an issue where some dependencies where reported with an inaccurate name.
CI Visibility
- Resolves an issue where JUnit XML output would not count tests retried by Early Flake Detection, Auto Test Retries, and Attempt-to-Fix.
Lib-Injection
- Avoids zombie process from telemetry sender on startup.
LLM Observability
- Resolves an issue where large spans traced within a short time interval were dropped despite being under the 1 MB limit.
- Resolves an issue with anthropic LLM spans where multiple system prompts caused missing input messages.
- Resolves an issue where traces containing intermixed APM and LLM Observability spans caused incorrect parent IDs for LLM Observability spans.
- Fixes an issue where LLMObs could not be enabled in a forked process when setting agentless_enabled=True or DD_LLMOBS_AGENTLESS_ENABLED=true.
- openai: Avoids creating spans for streamed Open AI chat and completion requests that use with_raw_response since the stream cannot be traced in these cases, leading to unfinished spans.
Profiling
- Resolves an issue where the Lock profiler would throw an AttributeError: '_ProfiledThreadingLock' object has no attribute '_self_acquired_at'.
Single-Step Instrumentation
- Resolves an issue where using Poetry to run an application under a version of Python different from the version used to install Poetry caused automatic library injection to fail.
Tracing
- graphql: Fixes an issue in GraphQL patching that threw an error when locations (an Optional field) is None. This checks for locations before setting attributes.

Other Changes

Single-Step Instrumentation
- Ensures that the SSI is not used for uWSGI applications. For enablement instructions, refer to the following our advanced_usage docs.

3.2.2

Bug Fixes

ASM
- Fixes a NotImplementedError that occurred when trying to deepcopy wrapped builtin functions (like open) while ASM or IAST were enabled. The error was caused by the wrapper not implementing the __deepcopy__ method.
- SCA: Resolves an issue where some dependencies where reported with an inaccurate name.
CI Visibility
- Resolves an issue where JUnit XML output would not count tests retried by Early Flake Detection, Auto Test Retries, and Attempt-to-Fix.
LLM Observability
- Resolves issue where Pydantic model objects were encoded as unhelpful default placeholder text.
- Improves encoding of non-JSON serializable I/O objects by attempting to convert to string before resorting to default placeholder text.
Profiling
- Resolves an issue where the Lock profiler would throw an AttributeError: '_ProfiledThreadingLock' object has no attribute '_self_acquired_at'.
Tracing
- pylibmc: Fixes an issue where using Client(server=[url]) would throw the error __init__() got multiple values for argument 'servers'

3.1.1

Bug Fixes

ASM
- SCA: Resolves an issue where some dependencies where reported with an inaccurate name.
CI Visibility
- Resolves an issue where ddtrace's own sys.monitoring coverage tool in Python 3.12+ would block other sys.monitoring tools such as pytest-cov from being used.
- Resolves an issue where JUnit XML output would not count tests retried by Early Flake Detection, Auto Test Retries, and Attempt-to-Fix.
LLM Observability
- Resolves issue where Pydantic model objects were encoded as unhelpful default placeholder text.
- Improves encoding of non-JSON serializable I/O objects by attempting to convert to string before resorting to default placeholder text.
Profiling
- Resolves an issue where the Lock profiler would throw an AttributeError: '_ProfiledThreadingLock' object has no attribute '_self_acquired_at'.
Tracing
- Changes the name of the span event generated by a record_exception call as it was not following OTEL semantics.

2.21.2

Bug Fixes

single-step instrumentation: Removes boto3 from the minimum versions list to avoid blocking SSI on boto3 version.
internal: Fix performance overhead of Python distribution parsing for internal telemetry.
ASM: This fix resolves an issue where blocking mechanism could partially fail with a 500 error on fastapi with python>=3.11 with a custom middleware.

3.3.0

New Features

Includes dynamodb in the default list of cloud services for payload tagging, i.e. DD_TRACE_CLOUD_PAYLOAD_TAGGING_SERVICES Note that cloud services payload tagging feature is still gated by DD_TRACE_CLOUD_REQUEST_PAYLOAD_TAGGING and DD_TRACE_CLOUD_RESPONSE_PAYLOAD_TAGGING.
LLM Observability: This introduces tracing for converse calls to the Bedrock Converse API. converse calls are traced as
LLM spans with max_tokens, temperature inference parameters and input/output messages.
botocore: This introduces capturing prompts, token usage, and inference parameters for converse calls to the Bedrock API.
SCA: This add support for extended heartbeat every 24h with dependencies payload.
kafka: Adds messaging.destination.name tag to kafka produce and consume spans.
LLM Observability: Avoids submitting spans to LLM Observability when a non-default base URL is detected in request for the Anthropic, Bedrock, LangChain, Open AI, and Azure Open AI integrations.
ASM: This introduces support for automatic instrumentation of session monitoring and blocking for Django.
ASM: This introduces new metrics span tags to keep track of truncations of WAF payloads.
runtime_metrics: Adds support for sending runtime metrics as gauge metrics (instead of distributions). To enable this feature set DD_TRACE_EXPERIMENTAL_FEATURES_ENABLED=DD_RUNTIME_METRICS_ENABLED.
runtime_metrics: Adds support for tagging runtime metrics with the current runtime ID. To enable tagging, set DD_TRACE_EXPERIMENTAL_RUNTIME_ID_ENABLED=True.

Bug Fixes

SCA: This fix resolves an issue where some dependencies where reported with an inaccurate name.
profiling: This fix resolves an issue where the Lock profiler would throw an AttributeError: '_ProfiledThreadingLock' object has no attribute '_self_acquired_at'.
tags: Updates DD_TAGS parsing to match the Datadog Agent and other SDKs. Key-value pairs can now be separated by both commas and spaces (e.g., DD_TAGS="key1:value1, key2:value2, key3:value3"). This reverts a change introduced in v2.11.0, which only allowed one separator type.
dynamic sampling: Ensures that dynamic sampling rates are always consistent with the rates received from the agent via remote configuration. Previously, setting a global sample rate and then sampling rules via remote config would result in both configurations being applied. With this fix only the most recent sampling configuration will be applied.
LLM Observability: This fix resolves an issue where enabling LLM Observability caused all botocore submodules to be patched instead
of only bedrock-runtime.
LLM Observability: Resolves issue where Pydantic model objects were encoded as unhelpful default placeholder text.
LLM Observability: Improves encoding of non-JSON serializable I/O objects by attempting to convert to string before resorting to default placeholder text.
tracing: Resolves an issue where trace information, such as span links, baggage, and trace-level tags (ex: sampling decision maker), could be lost when a new process was created while a trace was active.
internal: Fixes an issue where trimming a traceback to attach it to the span could result in the loss of the most recent frames.
pylibmc: fixes an issue where using Client(server=[url]) would throw the error __init__() got multiple values for argument 'servers'
tracing: Ensures remote sampling rules no longer overwrite existing samplers, preserving service-based sample rates set by the Agent and ensuring tracer rate limits configured by ddtrace products (e.g., ASM) are always respected.

Other Changes

tracing: Ensures a single DatadogSampler is initialized per Tracer instance, improving Tracer object initialization speed by approximately 7% and simplifying sampling logic.
agent: Migrates agent connection configurations to envier. This provides better documentation and validation for expected types. The following configuration are affected:
- DD_TRACE_AGENT_URL
- DD_DOGSTATSD_URL
- DD_TRACE_AGENT_HOSTNAME
- DD_AGENT_HOST
- DD_TRACE_AGENT_PORT
- DD_DOGSTATSD_PORT
- DD_AGENT_PORT
- DD_TRACE_AGENT_TIMEOUT_SECONDS Of the above impacted configurations, these had a precedence change:
- DD_TRACE_AGENT_HOSTNAME over DD_AGENT_HOST
- DD_TRACE_AGENT_PORT over DD_AGENT_PORT
- DD_DOGSTATSD_HOST over DD_AGENT_HOST
- DD_DOGSTATSD_PORT over DD_AGENT_PORT

2.21.4

Bug Fixes

CI Visibility: This fix resolves an issue where ddtrace's own sys.monitoring coverage tool in Python 3.12+ would block other sys.monitoring tools such as pytest-cov from being used.

3.2.0

New Features

ASM: This introduces auto instrumentation of signup events for Django.

Upgrade Notes

ASM: To align across products, we're replacing DD_EXPERIMENTAL_APPSEC_STANDALONE_ENABLED with DD_APM_TRACING_ENABLED. Feature set remains the same.
Bumps libdatadog dependency to v16.0.3.

Bug Fixes

celery: When multiple broker URLs are provided as a list, use the first broker URL from a list to avoid parsing errors.
CI Visibility: This fix resolves an issue where ddtrace's own sys.monitoring coverage tool in Python 3.12+ would block other sys.monitoring tools such as pytest-cov from being used.
tracing: Changes the name of the span event generated by a record_exception call as it was not following OTEL semantics.
Fixes an issue where Profiling native threads would respect the musl libc default stack size, which could cause stack overflows in certain configurations.
profiling: The memory profiler has a guard to avoid re-entering its code if there are allocations during sampling. This guard was meant to be thread-local, but was not correctly declared as such. This doesn't immediately cause problems because the profiler uses try-locks to protect access to its data structures, and re-entering the code will fail to acquire the locks. But this bug could be a source of deadlocks or data corruption if the code changes substantially in the future. This fix makes the guard thread-local as originally intended.

3.1.0

New Features

API Gateway: Add preview support to create inferred AWS API Gateway spans when certain AWS Gateway headers are passed into the request. Requires DD_TRACE_INFERRED_PROXY_SERVICES_ENABLED=true to enable. For enablement instructions, contact support.
LLM Observability: Properly labels Deepseek for the model provider and span name for LLM Observability spans for calls to Deepseek models using the OpenAI SDK.
tracing: Introduces a record_exception method that adds an exception to a Span as a span event. Refer to [Span.record_exception](https://ddtrace.readthedocs.io/en/stable/api.html#ddtrace.trace.Span.record_exception) for more details.
LLM Observability: Introduces an improved automated context management system for LLM Observability-specific spans. Also modifies LLMObs.export_span(), LLMObs.inject_distributed_headers(), LLMObs.annotate() to default to the current active LLM Observability-specific span if span is not provided.
LLM Observability: Introduces automatic distributed tracing support for LLM Observability traces involving the concurrent.futures.thread module.
Symbol Database is now enabled by default.
graphql: Add span events on GraphQL error spans, allowing multiple errors to be surfaced on the span.
graphql: Adds user-provided extensions to graphql error span events through the DD_TRACE_GRAPHQL_ERROR_EXTENSIONS environment variable. This is disabled by default; when set it allows users to capture their extensions.
LLM Observability: This introduces metadata argument to LLMObs.submit_evaluation_for, which enables users to
attach arbitrary structured metadata to custom evaluations.

Bug Fixes

tracing: Changes the name of the span event generated by a record_exception call as it was not following OTEL semantics.- internal: Require bytecode>=0.15.1 for Python 3.12 to avoid known bug.
asyncio: Fixes an issue where the name of a coroutine was being overridden by a ddtrace function.
single-step instrumentation: Removes boto3 from the minimum versions list to avoid blocking SSI on boto3 version.
configurations: This fix resolves an issue where DD_TRACE_CLOUD_PAYLOAD_TAGGING_SERVICES env variable was not parsed correctly
tracing: Captures the execution of async functions traced via DD_TRACE_METHODS, not just coroutine creation. This change increases span durations, which may affect latency-based metrics.
ASM: This ensures that no module from ASM are loaded when ASM is disabled or unavailable. SCA: This ensures that no module from IAST are loaded when IAST is disabled or unavailable.
Resolves ModuleNotFoundError errors raised when a OpenTracing tracer is initialized with a AsyncioScopeManager.
config: Resolves an issue where accessing ddtrace.config.<undefined_attribute> would create an IntegrationConfig object instead of raising an AttributeError.
botocore: This fix moves bedrock token usage metrics on APM spans from the meta field under bedrock.usage.{prompt/completion}_tokens to the metrics field under bedrock.response.usage.{prompt/completion}_tokens.
span: Fix issue where spans weren't being handled correctly and were not being sent when using a custom Exception class that raises an exception in __str__.
LLM Observability: This fix resolves an issue where the Gemini and Vertex AI integrations were not properly submitting spans for streamed responses in the case that the generator is not fully consumed.
internal: Fix ddtrace internal logger initialization mutating an unlocked shared resource.
internal: Fix performance overhead of Python distribution parsing for internal telemetry.
LLM Observability: This fix resolves an issue with correlating traces between APM and LLM Observability where trace IDs were always being converted to hexadecimal values which broke APM links for 64-bit trace IDs.
profiling: Fixes a crash in memory allocation profiler when tracking a large number of live allocations, caused by unsigned integer wraparound which lead to invalid memory access
pymongo: Changes the mongodb.query span tag to json so it can be parsed by the agent.
ASM: This fix resolves an issue where blocking mechanism could partially fail with a 500 error on fastapi with python>=3.11 with a custom middleware.
Code Security (IAST): In some applications, unique cookie names are used per request, leading to a high number of unique cookie vulnerabilities. This fix improves detection accuracy by identifying the vulnerability at the set_cookies function and adding more context. Warning: This change updates the vulnerability hash, meaning that if it was previously silenced in the web dashboard, it may reappear as a new vulnerability.
LLM Observability: Fixes non-LLM message inputs and outputs to be rendered as json rather than Python strings.
profiling: fixes an issue where the profiler unnecessarily outputs log messages when it was unable to record a lock acquire event.
baggage: Fixed an issue where ot-baggage headers in wsgi apps were not properly recognized due to the http_ header prefix. This ensures consistent baggage propagation in wsgi compatible web frameworks.
runtime_metrics: Ensures only the global service name is tagged on runtime metrics, instead of every service name found in the process.
single-step instrumentation: resolves an issue in which incorrect runtime version constraints prevented injection into processes using Python 3.13.1+
library: Updates the exception raised when ddtrace.patch(...) is called with an invalid argument.

2.21.3

Bug Fixes

configurations: This fix resolves an issue where DD_TRACE_CLOUD_PAYLOAD_TAGGING_SERVICES env variable was not parsed correctly

2.21.3

Bug Fixes

configurations: This fix resolves an issue where DD_TRACE_CLOUD_PAYLOAD_TAGGING_SERVICES env variable was not parsed correctly

3.0.0

Migration Guide: Upgrading from dd-trace-py v2 to v3

We are happy to announce the release of v3.0.0 of ddtrace. This release drops support for Python 3.7 and many previously-deprecated parts of the library interface.

The 2.x release line enters Maintenance Mode with the release of 3.0.0, meaning that it will only receive bugfix changes on its last few minor releases, and 2.21 will be the last 2.x minor version. See the versioning policy for support level definitions.

Summary of Significant Changes

Removed support for Python 3.7 from the library

CI Visibility

The new pytest plugin is now the default.
Module, suite, and test names are now parsed from item.nodeid.
Test names now include the class name (TestClass::test_method) for class-based tests.
Test skipping is now done at the suite level.
The DD_PYTEST_USE_NEW_PLUGIN_BETA environment variable is no longer used.

LLM Observability

Removed support for DD_LLMOBS_APP_NAME.
The parameters argument in LLMObs.annotate() has been removed. Use metadata instead.
Dropped support for OpenAI 0.x and Langchain v0.0.x.

Tracing

Removed all deprecated interfaces from integrations.
DD_TRACE_PROPAGATION_STYLE=b3 single header is no longer supported. Use b3 instead.
Dropped support for multiple Tracer instances. The global tracer instance ddtrace.tracer must be used.
Removed deprecated parameters from Tracer.configure(...).
Dropped support for several deprecated tracing configurations. See “Environment Variables” section below for details.

Recommended Pre-Upgrade Steps

Before upgrading to v3, check for deprecation warnings in your application using the latest v2 release (2.21.0):

Detect Deprecation Warnings in Tests

Enable warnings as errors in pytest:

pytest -W "error::ddtrace.DDTraceDeprecationWarning" tests.py

Detect Deprecation Warnings in Applications

Set the following environment variable to raise exceptions for deprecated usage:

DD_TRACE_RAISE_DEPRECATIONWARNING=1 ddtrace-run python app.py

Alternatively, enable all deprecation warnings:

PYTHONWARNINGS=all python app.py

Environment Variables

The following environment variables have been removed or replaced in v3. Use the table below to update your configuration:

v2	Removed in v3	Replaced/Changed in v3
LLM Observability
`DD_LLMOBS_APP_NAME`		Use `DD_LLMOBS_ML_APP` instead
`_DD_LLMOBS_EVALUATOR_SAMPLING_RULES`		Use `DD_LLMOBS_EVALUATOR_SAMPLING_RULES` instead
`_DD_LLMOBS_EVALUATORS`		Use `DD_LLMOBS_EVALUATORS` instead
Tracing
`DD_ANALYTICS_ENABLED`	No replacement, this setting is now a no-op. Datadog Analytics is no longer supported. See the ingestion controls documentation for more detail.
`DD_HTTP_CLIENT_TAG_QUERY_STRING`		Use `DD_TRACE_HTTP_CLIENT_TAG_QUERY_STRING`.
`DD_TRACE_ANALYTICS_ENABLED`	No replacement, this setting is now a no-op. Datadog Analytics is no longer supported. See the ingestion controls documentation for more detail.
`DD_TRACE_METHODS` (using `[]` notation)		Must use `:` notation instead, For example: `mod.submod:method1,method2;mod.submod:Class.method1`. (docs)
`DD_TRACE_PROPAGATION_STYLE=”b3 single header”`		Use `DD_TRACE_PROPAGATION_STYLE=b3` for identical behavior
`DD_TRACE_SAMPLE_RATE`		Use `DD_TRACE_SAMPLING_RULES` (docs).
`DD_TRACE_SPAN_AGGREGATOR_RLOCK`	No replacement (feature removed). Config variable is now a no-op.
CI Visibility
`DD_PYTEST_USE_NEW_PLUGIN_BETA`	No replacement (new pytest plugin is not in beta anymore).

Use the following patterns to identify the deprecated environment variables in a code base:

git grep -P -e "DD_LLMOBS_APP_NAME" \
  -e "_DD_LLMOBS_EVALUATOR_SAMPLING_RULES" \
  -e "_DD_LLMOBS_EVALUATORS" \
  -e "DD_TRACE_PROPAGATION_STYLE=.*b3 single header" \
  -e "DD_TRACE_SAMPLE_RATE" \
  -e "DD_TRACE_API_VERSION=v0.3" \
  -e "DD_ANALYTICS_ENABLED" \
  -e "DD_TRACE_ANALYTICS_ENABLED" \
  -e "DD_HTTP_CLIENT_TAG_QUERY_STRING" \
  -e "DD_TRACE_SPAN_AGGREGATOR_RLOCK" \
  -e "DD_TRACE_METHODS=.*\[\]"

Replace deprecated settings with the recommended alternatives where applicable.

NOTE: The changes to environment variables apply only to the configuration of the dd-trace-py library and not the Datadog Agent.

Legacy Interfaces

The following methods and module attributes have been removed or changed. Unless noted, removal means that the functionality pointed to by the removed interface has been disabled.

v2	v3
Tracing
`Tracer.configure( enabled=foo, hostname=foo, port=foo, uds_path=foo, https=foo, sampler=foo, wrap_executor=foo, priority_sampling=foo, settings=foo, dogstatsd_url=foo, writer=foo, partial_flush_enabled=foo, partial_flush_min_spans=foo, api_version=foo )`	`Tracer.configure()`
`ddtrace.opentracer.tracer.Tracer()._dd_tracer`	`ddtrace.tracer`
`ddtrace.tracer.sampler.rules[].choose_matcher(lambda: True)`	`ddtrace.tracer.sampler.rules[].choose_matcher(“foo”)` Callables and regex patterns are no longer allowed as arguments to `choose_matcher`
`Span.sampled`	`Span.context.sampling_priority > 0`
LLM Observability
`LLMObs.annotate(parameters=foo)`	`LLMObs.annotate(metadata=foo)`

Python 3.13 Support

Much of the library’s functionality has been compatible with Python 3.13 since version 2.20. Some pieces of functionality do not yet work with Python 3.13. Here’s what does not work under Python 3.13 as of dd-trace-py 3.0. 'Not tested' indicates that while the feature may be compatible with Python 3.13, its functionality has not been verified, as it is not included in our automated test coverage.

dd-trace-py doesn’t work on Windows with Python 3.13
Appsec Threat Detection is not tested against Django, Flask, or FastAPI with 3.13
Automatic Service Naming is not tested with 3.13
The following products are not tested with 3.13:
- Code Coverage
- Appsec IAST
- Data Streams Monitoring
- CI Visibility
- Continuous Profiling
The following integrations are not tested with 3.13:
- anthropic
- consul
- freezegun
- gevent
- google_generativeai
- gunicorn
- langchain
- mysqlclient
- opentracing
- psycopg
- psycopg2
- pymysql
- pytest
- pytest-bdd
- pytest-benchmark
- sanic
- selenium
- sqlite3
- starlette
- tornado
- vertexai

By following these guidelines, you can transition to dd-trace-py v3 gradually, ensuring minimal disruption to your services while taking advantage of the latest features and improvements. Happy migrating!

New Features

Removes code and tests related to Python 3.7, breaking the library's compatibility with this Python version.
ASM
- Introduces use of meta struct for appsec event reports. This will improve performances. ASM will now require at least agent version 7.35.0 to work as expected.
Code Security (IAST)
- Introduces Code Security support to PyMySQL and MySQLDB integrations.
- XSS detection for Django applications and Jinja2 (Flask and FastAPI applications), which will be displayed on your DataDog Vulnerability Explorer dashboard. See the Application Vulnerability Management documentation for more information about this feature.
- Implements the detection of the Stacktrace-Leak vulnerability for Django, Flask and FastAPI.
LLM Observability
- valkey: Adds automatic instrumentation of the Valkey package. Thank you AhmadMasry!
- openai: Introduces tracing support to the OpenAI integration for Python versions 3.12 and 3.13.
- Introduces an integration with the RAGAS evaluation framework to continuously monitor the performance of context-augmented LLM generations in production. The integration supports evaluating LLM inferences with the following RAGAS metrics:
  - Faithfulness: measures if the LLM response is faithful to the provided context.
  - Answer Relevancy: measures how relevant the LLM response is to the user input.
  - Context Precision: measures how effectively the context is used in the generated response.
  To learn more, see the LLM Observability evaluations guide.

Removed in v3

Replaced/Changed in v3

LLM Observability

DD_LLMOBS_APP_NAME

Use DD_LLMOBS_ML_APP instead

_DD_LLMOBS_EVALUATOR_SAMPLING_RULES

Use DD_LLMOBS_EVALUATOR_SAMPLING_RULES instead

_DD_LLMOBS_EVALUATORS

Use DD_LLMOBS_EVALUATORS instead

Tracing

DD_ANALYTICS_ENABLED

No replacement, this setting is now a no-op. Datadog Analytics is no longer supported. See the ingestion controls documentation for more detail.

DD_HTTP_CLIENT_TAG_QUERY_STRING

Use DD_TRACE_HTTP_CLIENT_TAG_QUERY_STRING.

DD_TRACE_ANALYTICS_ENABLED

No replacement, this setting is now a no-op. Datadog Analytics is no longer supported. See the ingestion controls documentation for more detail.

DD_TRACE_METHODS (using [] notation)

Must use : notation instead, For example: mod.submod:method1,method2;mod.submod:Class.method1. (docs)

DD_TRACE_PROPAGATION_STYLE=”b3 single header”

Use DD_TRACE_PROPAGATION_STYLE=b3 for identical behavior

DD_TRACE_SAMPLE_RATE

Use DD_TRACE_SAMPLING_RULES (docs).

DD_TRACE_SPAN_AGGREGATOR_RLOCK

No replacement (feature removed). Config variable is now a no-op.

CI Visibility

DD_PYTEST_USE_NEW_PLUGIN_BETA

No replacement (new pytest plugin is not in beta anymore).

git grep -P -e "DD_LLMOBS_APP_NAME" \ -e "_DD_LLMOBS_EVALUATOR_SAMPLING_RULES" \ -e "_DD_LLMOBS_EVALUATORS" \ -e "DD_TRACE_PROPAGATION_STYLE=.*b3 single header" \ -e "DD_TRACE_SAMPLE_RATE" \ -e "DD_TRACE_API_VERSION=v0.3" \ -e "DD_ANALYTICS_ENABLED" \ -e "DD_TRACE_ANALYTICS_ENABLED" \ -e "DD_HTTP_CLIENT_TAG_QUERY_STRING" \ -e "DD_TRACE_SPAN_AGGREGATOR_RLOCK" \ -e "DD_TRACE_METHODS=.*\[\]"

Tracing

Tracer.configure( enabled=foo, hostname=foo, port=foo, uds_path=foo, https=foo, sampler=foo, wrap_executor=foo, priority_sampling=foo, settings=foo, dogstatsd_url=foo, writer=foo, partial_flush_enabled=foo, partial_flush_min_spans=foo, api_version=foo )

Tracer.configure()

ddtrace.opentracer.tracer.Tracer()._dd_tracer

ddtrace.tracer

ddtrace.tracer.sampler.rules[].choose_matcher(lambda: True)

ddtrace.tracer.sampler.rules[].choose_matcher(“foo”) Callables and regex patterns are no longer allowed as arguments to choose_matcher

Span.sampled

Span.context.sampling_priority > 0

LLM Observability

LLMObs.annotate(parameters=foo)

LLMObs.annotate(metadata=foo)