Python agent release notesRSS

Notes

This release of the Python agent provides fixes for early activation in Django instrumentation, server side overrides for max_samples_stored being ignored, and guards against crashes from trace_cache.pop().

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

Bug fixes

• Prevents early activation in Django instrumentation

  • Remove calls to application_settings() in Django instrumentation to prevent issues with early activations in pre-fork applications.

• Fix server configuration to local configuration sync issues

  • This fixes a bug created when event_harvest_config.harvest_limits.* settings were deprecated in favor of *.max_samples_stored settings. Some server side configurations were not overriding the client side configurations, resulting in data being sent to the collector despite the user disabling this.

• Guard against crashes from trace_cache.pop()

  • In the ContextOf class, there was an unsafe access of the trace_cache via pop(key) without a default value. This caused crashes when the weakref in the dictionary was already garbage collected.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent makes the following changes:

• Removes support for Python 3.7
• Removes and or deprecates various APIs and settings
• Deprecates aioredis instrumentation
• Raises the minimum supported version of setuptools and setuptools_scm
• Adds new instrumentation for AutoGen
• Adds new instrumentation for Pyzeebe
• Adds new MCP (Model Context Protocol) named spans
• Fixes a crash in psycopg
• Ensures MCP spans are only recorded when AI monitoring is enabled

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

Removed features

• Removes support for Python 3.7

  • Python 3.7 was EOL'd June 27, 2023. The Python agent has dropped support for it.

• Removes various deprecated APIs

  • The following APIs have been removed:

    Legacy API	Replacement API
    accept_distributed_trace_payload	accept_distributed_trace_headers
    add_custom_parameter	add_custom_attribute
    add_custom_parameters	add_custom_attributes
    create_distributed_trace_payload	insert_distributed_trace_headers
    get_browser_timing_footer	The functionality of this has been wrapped into get_browser_timing_header()
    record_exception	notice_error()
    NewRelicLogHandler	NewRelicLogForwardingHandler
    lambda_handler & LambdaHandlerWrapper	These APIs have moved to the newrelic-lambda package.

• Removes various deprecated settings

  • The following settings have been removed:

    Legacy Setting	Replacement Setting
    transaction_tracer.capture_attributes	transaction_tracer.attributes.enabled
    error_collector.capture_attributes	error_collector.attributes.enabled
    browser_monitoring.capture_attributes	browser_monitoring.attributes.enabled
    analytics_events.enabled	transaction_events.enabled
    analytics_events.capture_attributes	transaction_events.attributes.enabled
    error_collector.ignore_errors	error_collector.ignore_classes
    strip_exception_messages.whitelist	strip_exception_messages.allowlist
    ssl	This is now hardcoded to true so it has been removed as a setting option.
    ignored_params	Use newrelic.attributes.exclude = request.parameters.{name} instead. Add each request parameter key to the newrelic.attributes.exclude list, delimited by commas. Be sure to prepend the key with request.parameters.
    agent_limits.merge_stats_maximum	This is now completely ignored in favor of serverside config.

Deprecated features

• Instrumentation for aioredis has been deprecated

  • The aioredis package has been deprecated in favor of redis which now includes the submodule redis.asyncio. As a result, the instrumentation for aioredis has also been deprecated. Please use the redis library instead. This instrumentation will be removed in a future release.

• Deprecates some existing harvest settings to match other language agents

  • In order to be more consistent with the conventions of the other language agents, the following harvest settings are now deprecated:

    Legacy Setting	Replacement Setting
    event_harvest_config.harvest_limits.analytic_event_data	transaction_events.max_samples_stored
    event_harvest_config.harvest_limits.span_event_data	span_events.max_samples_stored
    event_harvest_config.harvest_limits.error_event_data	error_collector.max_event_samples_stored
    event_harvest_config.harvest_limits.custom_event_data	custom_insights_events.max_samples_stored
    event_harvest_config.harvest_limits.log_event_data	application_logging.forwarding.max_samples_stored

    This is a revert of a previous change that renamed these settings from *.max_samples_stored to the event_harvest_config.harvest_limits.* form. This decision has been reversed, and the original settings named have been restored to match the other language agents who did not change these settings names. The deprecated settings will continue to work for now, but they will be removed in a future release.

• Deprecates configuration of the agent through the WSGI environ dictionary

  • The following settings which were previously possible to configure through the WSGI environ dictionary are now deprecated. Please configure these settings through environment variables or the config file.

    • newrelic.set_background_task
    • newrelic.suppress_apdex_metric
    • newrelic.suppress_transaction_trace
    • newrelic.capture_request_params
    • newrelic.disable_browser_autorum

• Deprecates support for building the agent with setuptools<61.2 or setuptools_scm<6.4

  • The minimum supported versions of setuptools and setuptools_scm have been increased to 61.2 and 6.4 respectively. This is to support the use of pyproject.toml for configuring the package metadata and building instructions. The agent can now be built with more modern build tools which will no longer log warnings about the lack of a pyproject.toml file.
  • This change adds new version attributes to the packages, newrelic.__version__ and newrelic.__version_tuple__, which can be used to programmatically check the version of the agent at runtime.

New features

• Adds new instrumentation for AutoGen

  • Adds instrumentation (named spans and custom Llm events) for agents and tools in AutoGen

• Adds new instrumentation for Pyzeebe

  • Adds automatic instrumentation and framework-specific agent attributes for Pyzeebe

• Adds new MCP (Model Context Protocol) named spans

  • Adds new named spans for MCP (Model Context Protocol) tool calls made via tool manager to support Streamable HTTP

Bug fixes

• Fixes a crash in psycopg>=3.2.0 when calling as_string()

  • Fixes an issue in psycopg>=3.2.0 where the as_string() method no longer required arguments, causing a potential crash. This has been fixed, and both signatures of this method are supported.

• Ensures MCP (Model Context Protocol) spans are only recorded when AI monitoring is enabled

  • Ensures MCP (Model Context Protocol) spans are only recorded when the setting ai_monitoring.enabled is enabled.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for the following:

• Custom tasks in Celery
• Graphene-Django monitoring
• Middleware filtering in Django
• AWS Bedrock Converse API
• W3CTraceParent header sampled flag

This release also provides the following fixes:

• Safeguards for the deepest unique path in GraphQL
• Safeguards for web URL request parsing
• Fallback for ElasticSearch and GraphQL when trace.settings==None
• Remove options method from async ElasticSearch client list

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Adds support for custom tasks in Celery

  • Add monitoring support for custom task classes in Celery. Previously, only tasks using the default task class were able to be monitored. Note: This version of the Python agent will no longer support Celery versions less than v5.2.0.

• Adds support for Graphene-Django

  • Add monitoring support for synchronous schemas created with Graphene-Django. Previously, schemas created with Graphene-Django would only be partially monitored, resulting in missing GraphQL related attributes.

• Adds support for middleware filtering in Django

  • Users are now able to filter which Django middleware to monitor. There are now three additional settings:

    • instrumentation.middleware.django.enabled (Default is true)
    • instrumentation.middleware.django.exclude
    • instrumentation.middleware.django.include

  See the Django Middleware Filtering page for more information and examples.

• Adds support for AWS Bedrock Converse API

  • Add monitoring support for non-streaming Converse API. This includes support for converse chat completion calls made using boto3 and aioboto3 clients.

• Adds support for W3CTraceParent header sampled flag

  • When an upstream service sends the W3CTraceParent header with the sampling flag set, New Relic will take into account that sampling decision when deciding whether the current transaction will be sampled.

• Added support for sampling decisions based on whether the remote parent is sampled

  • Added support for two new configuration options:

    • distributed_tracing.sampler.remote_parent_sampled
    • distributed_tracing.sampler.remote_parent_not_sampled

  • default is the default behavior and behaves as it did before this change (it passes the sampling decision to the adaptive sampling algorithm to determine whether the transaction will be sampled).

  • always_on means that the trace that has a remote parent sampled or remote parent not sampled will always be sampled.

  • always_off means that the trace that has a remote parent sampled or remote parent not sampled will never be sampled.

Bug fixes

• Add safeguard for deepest unique path in GraphQL

  • Some users were experiencing issues with the deepest unique path in GraphQL not being generated due to a field name returning None. This issue has been fixed.

• Add safeguard for web request parsing

  • Some users experienced a parsing error when request_uri was malformed or contained invalid characters. This can happen at this point if the request_uri in the request header is invalid and does not match the (valid) URL originally given. While this safeguard has now been put in place in New Relic, urllib has resolved this issue for versions of Python released after Jan 31, 2025.

• Fix crash in ElasticSearch and GraphQL when trace.settings==None

  • In cases where trace.settings is None, the instrumentation for ElasticSearch crashed the application. A global_settings fallback has been put in place. This also resolves the same issue found in GraphQL.

• Remove options method from async ElasticSearch client list

  • The previous instrumentation of ElasticSearch caused a crash because options was in the list of async methods. This method is no longer wrapped in a DatastoreTrace since it is not one that conducts any datastore operations/requests.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for new Redis functions, convert all internal paths from os.path to pathlib.Path, and fixes an issue with JSON encoding PosixPaths.

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Adds support for new Redis functions

  • Add instrumentation for new Redis functions.

• Convert all internal paths from os.path to pathlib.Path

  • Convert all internal usage of string based paths and os.path to pathlib.Path in preparation to support Windows paths.

Bug fixes

• Fix issue with JSON encoding PosixPath

  • Some users were experiencing issues with JSON encoding PosixPath configuration settings. This issue has been fixed.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for oracledb, fixes a bug in Kombu and Kafka where MessageTransaction object has no attribute destination_name, and fixes the disappearance of utilization metadata with AWS.

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Adds support for oracledb

  • Add instrumentation for oracledb.

Bug fixes

• Fix "MessageTransaction object has no attribute destination_name" bug

  • Fix a bug in MessageBroker instrumentations where the serialize methods were looking for the destination_name attribute on the MessageTransaction and it did not exist.

• Fix AWS utilization metadata issue when running AWS and K8s together

  • Due to permissions settings in K8s pods running on AWS EC2 instances, the utilization metadata is able to be retrieved the first time. However, if the application is left running continuously, the agent will reconnect and the utilization metadata fails to load. In cases like these, the utilization metadata is cached per agent instance and is able to be retrieved.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for Azure Function Apps and protobuf v6.

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Adds support for Azure Function Apps

  • Adds support for Azure Function Apps for HTTPTrigger types.

Bug fixes

• Adds pb2 files to enable protobuf v6 support

  • Previously, several google packages were not compatible with New Relic due to the lack of protobuf v6 support. This has now been resolved.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for the Model Context Protocol (MCP) and fixes import logic for importlib.metadata and pkg_resources.

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Add support for Model Context Protocol (MCP)

  • Adds support for MCP calls in mcp and fastmcp. The agent will now automatically instrument the following core MCP primitives:
    • Tools (via call_tool)
    • Resources (via read_resource)
    • Prompts (via get_prompt)

Bug fixes

• Fix import logic for importlib.metadata and pkg_resources

  • Updates logic around import of importlib.metadata to also attempt to use the importlib_metadata backport before falling back to pkg_resources.*. This also removes use of sys.version conditionals and fixes issues where Python 3.9 could only use pkg_resources despite having importlib.metadata available.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for Redis v6.0.0 and fixes conflicts with the opentelemetry-proto package.

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Add support for Redis v6.0.0

  • Adds support for all client methods introduced in Redis v6.0.0.

Bug fixes

• Fix conflicts with opentelemetry-proto package

  • Fixes an issue where the agent's internal copy of opentelemetry-proto would conflict with any other packages relying on opentelemetry-proto (such as ChromaDB). This should now be treated as a separate package by OpenTelemetry and prevent any conflicts.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.

Notes

This release of the Python agent adds support for Google Gemini via the Google Generative AI SDK and instruments new Kinesis methods.

Install the agent using easy_install/pip/distribute via the Python Package Index or download it directly from the New Relic download site.

New features

• Add support for Google Gemini via Google Generative AI SDK

  • Adds support for google-genai. The agent will automatically instrument embed_content calls for synchronous and asynchronous embeddings. The agent also now supports synchronous and asynchronous text generations for text inputs in non-streaming cases. This includes calls made to generate_content for single text-only inputs and calls to send_message for multi-turn conversations.

• Instrument new AWS Kinesis methods

  • Adds botocore instrumentation to support new AWS Kinesis methods including tag_resource, untag_resource, and list_tags_for_resource.

Support statement

We recommend updating to the latest agent version as soon as it's available. If you can't upgrade to the latest version, update your agents to a version no more than 90 days old. Read more about keeping agents up to date.

See the New Relic Python agent EOL policy for information about agent releases and support dates.