Python agent release notesRSS

Notes

This release of the Python agent fixes a memory leak that occurs on Python 2 when monitoring WSGI applications.

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 memory leak on Python 2 WSGI applications

  When using the 4.16.0 agent on Python 2 to monitor WSGI applications, the agent would cause memory to be allocated on each request. On Python 2, this memory was never deallocated due to a reference cycle. This has now been corrected.

Notes

This release of the Python agent adds support for native coroutines and generators with a new web_transaction API, as well as bugfixes.

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.

Features

• Add a new web_transaction API

  This method is now exposed both as a public API through newrelic.agent.api.wrap_web_transaction, and as a decorator with newrelic.agent.api.web_transaction. It can be used to instrument non-WSGI web transactions. For usage of this API see https://docs.newrelic.com/docs/agents/python-agent/python-agent-api/webt...

Bug Fixes

• The presence of unicode values in a SQL query on Python 2 may cause an application crash

  The agent was not properly handling unicode values when recording SQL queries in Python 2. As a result, the presence of unicode values in SQL queries in Python 2 applications may have resulted in an application crash. Unicode values in Python 2 SQL queries are now properly handled.

• The Python Agent may have improperly biased the random sampling of distributed traces

  Distributed traces are randomly sampled across services. The agent is responsible for propagating the state of its sampling decision to other services. In propagating that decision, an improper sampling bias may have been introduced in downstream services.

• The agent will no longer report the command arguments used to start newrelic-admin

  The agent previously reported the startup command and arguments when using newrelic-admin run-program and newrelic-admin run-python. The command and arguments are no longer sent to New Relic. This data may continue to be logged to the agent log files when debug level logs are enabled.

Notes

This release of the Python agent adds support for native coroutines and generators with the background task API and adds support for Kubernetes detection.

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.

Features

• @background_task can now be used with coroutines

  The background_task decorator API can now be used with native coroutines and generators.

  @background_task(name='my_coroutine')
   async def my_coroutine():
     await asyncio.sleep(0.1)

  Copia

• Add support for collection of Kubernetes metadata

  The agent will now collect environment variables prefixed by NEW_RELIC_METADATA_ as additional metadata. Some of this metadata may be added to Transaction events to provide context between your Kubernetes cluster and your services. For details on the benefits (currently in beta) see this blog post.

  The agent now also collects the KUBERNETES_SERVICE_HOST environment variable to detect when the application is running on Kubernetes. The data is used to link the application to the host machine when using the New Relic infrastructure product.

Bug Fixes

• Using time trace decorators allowed coroutines to become Iterables and generators to become awaitables

  When using time trace decorators native coroutines were converted to iterables. Additionally generators were converted to awaitables. The behavior of generators, native coroutines, and awaitable generators is now unchanged when using time trace decorators.

Notes

This release of the Python agent extends support of Amazon's boto3 library and includes bug fixes.

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.

Features

• AWS operation and request ID will now be reported in transaction traces and spans when using boto3 and botocore.

  The agent will now report aws.requestId and aws.operation for all calls to AWS made using botocore and boto3.

• DynamoDB calls are now reported under the Databases tab.

  The agent will now record DynamoDB query performance in the Databases tab in APM in addition to table name for the following calls:

  • put_item
  • get_item
  • update_item
  • delete_item
  • create_table
  • delete_table
  • query
  • scan

• Certain SQS calls will now report additional data for spans and transaction traces.

  The agent will now record the queue name in spans and transaction traces for the following SQS calls:

  • send_message
  • send_message_batch
  • receive_message

• SNS publish will now report additional data for spans and transaction traces.

  The SNS topic, target, or the string literal PhoneNumber will be reported to New Relic inside of spans and transaction traces.

• The full URL path will now be recorded on span events and transaction traces when using boto3 or botocore.

  The agent will now record the full URL path for API calls made to AWS through the boto3 / botocore libraries. The path will be available through span events and transaction traces.

Bug Fixes

• Using newrelic-admin to start a GunicornWebWorker with an aiohttp application factory resulted in an application crash.

  The agent would fail to start if using the newrelic-admin command to start an aiohttp application factory with GunicornWebWorker. This issue has now been fixed.

Notes

This release of the Python agent adds support for excluding attributes from span events and transaction trace segments, adds a new api called add_custom_parameters, and includes various bug fixes.

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.

Features

• Add ability to exclude attributes from span events and transaction trace segments

  This release adds support to exclude attributes (such as http.url) from span events (via the span_events.include/span_events.exclude options) and from transaction segments (via the transaction_segments.include/transaction_segments.exclude option).

  As with other attribute destinations, these new options will inherit values from the top-level attributes.include/attributes.exclude settings. See the documentation for more information.

• Add public add_custom_parameters API

  The method add_custom_parameters on Transaction is now exposed through newrelic.agent.add_custom_parameters

Bug Fixes

• Transaction counts were not reported for aiohttp's built-in error pages

  When a built-in error route was reached in aiohttp (such as a 404 due to a missing route), transactions were not recorded. As a result, the transaction counts may have been artificially low. aiohttp system route traffic will now be reported.

• aiohttp cross application tracing linking to non-Python applications may have been omitted if using multidict<3.0

  For aiohttp users using multidict versions less than 3.0, cross application tracing HTTP headers may have been generated in a way that was incompatible with non-Python applications. Headers are now generated in a format compatible with all New Relic agents.

• aiohttp 3.5.x versions generated agent instrumentation errors

  The agent previously failed to instrument aiohttp applications running versions 3.5.0 and greater. The agent now supports aiohttp versions up to 3.5.1.

Notes

This release of the Python agent adds support for Falcon framework, and includes various bug fixes.

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.

Features

• Add support for Falcon web framework

  This release adds support for the Falcon web framework. Data will now be automatically collected for applications using Falcon framework. The data will appear in both APM and Insights and will include performance details as well as information on application errors.

Bug Fixes

• "newrelic-admin record_deploy" now functions with proxies.

  The newrelic-admin recorddeploy command previously did not function when a proxy was defined by the newrelic.ini configuration file or the `NEW_RELIC_PROXY*` environment variables. This bug has now been fixed.

• Cross Application Tracing HTTP response headers were inserted on a 304 response

  When cross application tracing is enabled and the agent received a HTTP request from an application utilizing cross application tracing, the agent may have inserted additional response headers on a 304 HTTP response. The agent will no longer insert headers on a 304 response.

Notes

This release of the Python agent includes changes to the agent to enable monitoring of Lambda functions, improves the built-in Sanic instrumentation, and includes various bug fixes.

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.

Features

• Monitoring of Lambda functions

  This release includes changes to the agent to enable monitoring of Lambda functions. If you are interested in learning more or previewing New Relic Lambda monitoring please email lambda_preview@newrelic.com.

• Improve naming of Sanic HTTPMethodView view handlers

  Sanic views that were defined using the HTTPMethodView class were previously all named HTTPMethodView.as_view.<locals>.view regardless of the actual class in use. The agent will now name transactions after the actual view handler class.

Bug Fixes

• Fix ignored error reporting in CherryPy instrumention

  When missing query parameters, unexpected query parameters, unexpected positional arguments, or duplicate arguments were present in the CherryPy framework, a TypeError exception was recorded even when an ignored response status code (such as a 404) was generated. An error is no longer recorded when it results in the generation of an ignored status code.

• Excluding request.uri from transaction trace attributes hides it in the UI

  When request.uri is added to either attributes.exclude or transaction_tracer.attributes.exclude, the value will now no longer be sent to New Relic.

• Ability to disable sending request.uri as part of error traces

  When request.uri is added to either attributes.exclude or error_collector.attributes.exclude, the value will now no longer be sent to New Relic.

• Fix tracing of functions returning generators

  When tracing generators whose parent traces have ended an error was seen in the logs "Transaction ended but current_node is not Sentinel." This has now been fixed.

Notes

This release of the Python agent contains bug fixes.

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

• The creation of sampled events sometimes raised an exception in Python 3

  When more events (Transaction, Transaction Error, Custom, or Span) were created than allowed per harvest period in Python 3, sometimes a TypeError: '<' not supported between instances of 'dict' and 'dict' was raised. This issue has now been fixed.

Notes

This release of the Python agent introduces instrumentation for Sanic, and contains bug fixes.

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.

Features

• Add instrumentation for Sanic framework

  Data is now automatically collected for applications using the Sanic framework. Data for Sanic applications will appear in both APM and Insights. Additionally, cross application tracing and distributed tracing is supported for incoming requests for Sanic applications. In addition to service maps, Sanic applications will now show the calling application in transaction traces. See the instrumented packages page for more details.

Bug Fixes

• Explain plans were not generated when using psycopg2 named cursors

  When using named cursors in psycopg2, the agent attempted to generate an explain plan using the same named cursor. This resulted in a syntax error when the query was issued to the database. When using the default connection and cursor factories, the agent will now execute the explain query using only unnamed cursors.

• Convert bytes-like SQL statements to strings before obfuscating

  If a bytes-like object is used instead of a string when making a SQL call, a traceback was seen in the logs with TypeError: cannot use a string pattern on a bytes-like object. This issue has now been fixed.

• Save settings to MessageTrace objects

  If an external call using an instrumented http external library (for example requests) was used within a MessageTrace, a traceback was seen in the logs with AttributeError: 'MessageTrace' object has no attribute 'settings'. This issue has now been fixed.