Python agent release notesRSS

Notes

This release of the Python agent adds new instrumentation for the GraphQL frameworks Strawberry and Ariadne, updates to Graphene and Starlette GraphQL instrumentation, a change to logging configuration, and includes bug fixes.

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 new instrumentation for Strawberry

  GraphQL applications built using Strawberry will now be automatically instrumented.

• Add new instrumentation for Ariadne

  GraphQL applications built using Ariadne will now be automatically instrumented.

Improvements

• Add new and changed metrics for Graphene applications

  Metrics specifying the framework details for Graphene have been added, and GraphQL metrics now include Graphene as a product field.

  (eg. GraphQL/operation/Graphene/query/<query_name>/<query_path>)

• Improve tracing in Starlette GraphQL applications

  GraphQL applications using Starlette now support all executor classes. Previously, only async executor classes were supported.

Changes

• New Relic agent logs no longer propagate to the root logger by default

  Previously, logs using the logging module sent to the newrelic logger would propagate to the root logger, potentially causing duplicate logs to appear or be sent to New Relic. This has been changed to no longer propagate to the root logger by default.

  To reverse this behavior, in your application you can run logging.getLogger("newrelic").propagate = True at any point after initializing the agent. If you use the newrelic-admin wrapper, this can be done at any time inside your application.

Bug Fixes

• TimeTrace.str recursion

  A case of infinite recursion with TimeTrace.__str__ has been corrected in this release. Calling str(), repr() or print() on a TimeTrace object or subclass instance will now display helpful debugging information about it.

Notes

This release of the Python agent updates the log file location in the default newrelic.ini file, and includes multiple bug fixes.

The agent can be installed using easy_install/pip/distribute via the Python Package Index, or it can be downloaded directly from the New Relic download site.

Improved Features

• Update log_file location in default newrelic.ini

  The newrelic.ini generated using newrelic-admin generate-config has been updated to set the default log_file location to stdout for improved initial visibility. This setting can still be customized to log to stderr or a user-specified file as well.

Bug Fixes

• Handle cherryPy.HTTPRedirects

  cherrypy.HttpRedirects were incorrectly being reported as errors. This has now been corrected thanks to a contribution from @bdeeney!

• Address DeprecationWarning for invalid escape sequence

  A deprecation warning concerning unrecognized escape sequences in Python 3.6 (and higher) has been fixed. Thank you @urianchang for your contribution!

Notes

This release of the Python agent adds ARM64 wheels and fixes a crash in unsupported GraphQL versions.

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 wheel support for ARM64 (a.k.a. AArch 64, ARMv8)

  The agent now provides prebuilt wheels for ARM64 for Python versions >=3.6. This requires a version of pip>=19.3.

Bug Fixes

• Fix a crash in unsupported GraphQL versions

  A crash in GraphQL could occur when using an older version not supported by our instrumentation (ie. graphql-core<2.1). This has been corrected.

Notes

This release of the Python agent adds new GraphQL instrumentation support for the Graphene library and Graphene schemas created in Starlette/ FastAPI.

The agent can be installed using easy_install/pip/distribute via the Python Package Index or can be downloaded directly from the New Relic download site.

New Features

• Add support for Graphene

  Graphene applications will now be automatically instrumented.

• Add support for Graphene schemas in Starlette/FastAPI with use of an AsyncioExecutor

  Starlette/ FastAPI applications using GraphQL in conjunction with an AsyncioExecutor will now be automatically instrumented.

Notes

This release of the Python agent adds wildcarding to instrumentation config and fixes a crash in the memory metric sampler.

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

Improvements

• Add wildcard matching to instrumentation via config file

  The agent now supports wildcards for classes and functions when instrumenting via config file. Note: this does not include module paths.

Bug Fixes

• Fix a crash in memory sampling on BSD kernels

  A crash in the memory metric sampler was possible when using kernels besides MacOS and Linux. This has been corrected.

Notes

This release of the Python agent includes a bug fix related to middleware and background tasks in Starlette/ FastAPI.

The agent can be installed using easy_install/pip/distribute via the Python Package Index or can be downloaded directly from the New Relic download site.

Bug Fixes

• Fix Starlette/ FastAPI transaction naming and classification

  Starlette/ FastAPI routes that contained middleware and background tasks were being incorrectly classified as non-web transactions and named after the background task rather than the route. This has been corrected in this release.

Notes

This release of the Python agent updates Sanic instrumentation to support all available versions.

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

Improvements

• Update Sanic instrumentation

  The agent has updated support for all Sanic versions, with testing done on all versions >=18.12.0. Additionally, blueprint middleware is now supported.

• Update Pyramid transaction naming

  Previously, transaction names from exception views overrode route names. We've reversed this priority to be more in line with other instrumentation.

Bug Fixes

• Fix inspect.formatargspec deprecation warning

  A deprecation warning concerning inspect.formatargspec has been fixed.

• Fix HTTPX crash

  Initializing the agent after creating an HTTPX client could cause a crash in HTTPX. This has been corrected.

Notes

This release of the Python agent adds support for Flask v2, improves logging for HTTP exceptions, and includes a bug fix.

The agent can be installed using easy_install/pip/distribute via the Python Package Index or can be downloaded directly from the New Relic download site.

New Features

• Add support for Flask v2

  The agent will automatically instrument error handlers, nested blueprints, and async views.

Improvements

• Improve logging for 410 status codes

  The agent now logs the content of a data collector response which specifies the reason for a disconnect for 410 status codes.

Bug Fixes

• Reintroduce non-PID specific memory metrics

  In agent version v6.2.0.156, memory metrics were modified to add the PID to each metric name. This release reintroduces non-PID specific memory metrics.

Notes

This release of the Python agent deprecates the old record_exception API, replaces the error_collector.ignore_errors setting, adds a new notice_error API, adds expected errors functionality, improves the NewRelicContextFormatter, and fixes multiple bugs including a few potential crashes.

The agent can be installed using easy_install/pip/distribute via the Python Package Index or can be downloaded directly from the New Relic download site.

Deprecations

• Deprecated record_exception API

  record_exception has been superseded by the new notice_error API. The old API will forward exceptions to notice_error with a deprecation warning.

• Deprecated error_collector.ignore_errors setting

  error_collector.ignore_errors has been replaced with error_collector.ignore_classes to match other agents. Please update your config files to use the new setting. The old setting will continue to function with a deprecation warning.

New Features

• Added new notice_error API

  This new API falls much more in line with other language agents.

  It also includes expected errors functionality and a status code parameter.

  To see all changes, view the documentation for this API.

• Added new expected errors feature

  This feature implements expected errors as documented here. To enable them, use the following new settings.

  • error_collector.expected_classes
  • error_collector.expected_status_codes

Improved Features

• Improved NewRelicContextFormatter

  Adds new log_record_to_dict function to separate logic for JSON serialization and converting to a dictionary, enabling custom formatters to access the data more easily.

  Custom attributes now attempt JSON serialization by default and fallback to strings, meaning dictionaries and other serializable types are now supported.

  Thank you to @chrislawlor and @qvik-olli for your contributions!

• error_collector.ignore_status_codes is now configurable in server-side configuration

  Previously this setting was only available in the config file, but is now available via server-side configuration.

Bug Fixes

• Infinite tracing crash fix

  Crashes had occurred in infinite tracing following the inclusion of some startup timeout changes in v6.0.0. These changes have been rolled back.

• RuntimeInstrumentationError crash removed

  Crashes occurred when a RuntimeInstrumentationError was reached, indicating a potential problem with existing instrumentation. The exception is now caught and replaced with error level log messages to avoid crashing applications.

• SQLite instrumentation consumes generator objects passed to executemany

  A generator passed to Connection.executemany would unexpectedly be consumed by our instrumentation. This has been corrected and generators are now properly supported.

Known Issues

• Flask v2.0.0 support

  Flask released a new major version recently that we do not yet support, but expect to in the near future.

• Sanic support

  Sanic has released large changes to the framework recently that have broken some of our instrumentation on certain versions, and can crash the agent on others. We expect to correct these issues in the near future.