Python agent release notesRSS

Notes

This release of the Python agent 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.

Bug Fixes

• Fix a crash when using future-based interfaces on Tornado's http client

  Tornado's http client returns a future when fetch is called. Prior to this release, the agent caused a coroutine object to be returned rather than a future, causing a potential crash when application code assumed a future would be returned. A future will now be returned from http client fetch as expected.

• Fix a crash when running a coroutine created in another transaction

  When running a coroutine outside of the transaction that created it, the agent may have crashed if a separate transaction was active.

Notes

This release of the Python agent includes a bug fix for tornado and support for recording deployment markers in the EU region as well as specifying a deploy timestamp.

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

• Support for Real Time Streaming

  • Event data is now sent to New Relic every five seconds, instead of every minute. As a result, transaction, error, and custom events will now be available in New Relic and Insights dashboards in near real time. For more information on how to view your events with a five-second refresh, see the real time streaming documentation.
  • Note that the overall limits on how many events can be sent per minute have not changed. Also, span events, metrics, and trace data is unaffected, and will still be sent every minute.

• Added timestamp and EU region support for recording deployment markers

  When using the newrelic-admin record-deploy script an optional timestamp can be provided to specify when the deployment marker should be created. Furthermore support for accounts in the EU region have been added in this release.

Bug Fixes

• Fixed a crash in tornado versions >= 6.1

  Tornado 6.1 internals changed in a way that may have caused the agent's built-in instrumentation to fail. The instrumentation has been updated to alleviate this issue.

Notes

This release of the Python agent includes bug fixes for feedparser and django-piston.

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

• Fixed a crash when using feedparser

  When running with the feedparser library the agent would crash on attempting to parse an RSS feed. This issue has now been corrected.

• Fixed a crash when using django-piston

  When running with the django-piston library the agent would crash on attempting to service a request. This issue has now been corrected.

Notes

This release of the Python agent includes a bug fix for the mako library instrumentation.

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

• Fixed a crash when using the Mako template library

  When running with the Mako template library the agent would crash on attempting to render a template. This issue has now been corrected.

Notes

This release of the Python agent introduces:

• Event loop diagnostics feature
• Tornado instrumentation for versions 6 or higher
• Improved context manager API functionality and interfaces

This release also drops support for Python 3.4.

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.

Event loop diagnostics

Introducing event loop diagnostics for asyncio! The Python agent can now surface information about transactions that block the event loop. The agent will be able to generate information about transactions that have waited a significant amount of time to acquire control of the event loop.

Event loop diagnostic information will be available in both APM and via NRQL query.

Event loop metrics will appear in the transaction Breakdown table:

Time spent waiting on other transactions will be shown in the transaction Trace details page, as shown here:

eventLoopTime and eventLoopWait attributes will be available via NRQL query. An example query might be:
SELECT count(*) as 'count', average(eventLoopTime) as 'loopTime', average(eventLoopWait) as 'loopWait' FROM Transaction facet name

In New Relic Insights, this might show a result like this:

Diagnostics will be available by default through our existing aiohttp, sanic, and Tornado instrumentation. Diagnostic information will also be available for coroutines that use our background_task and web_transaction decorator APIs.

Tornado 6 Instrumentation

Instrumentation for Tornado framework version 6.x is supported without any feature flags. Support for versions earlier than Tornado 6.x has been dropped. The Tornado web server, framework, and http client are instrumented and information will show up in both APM and insights.

Important: Using the tornado.gen.coroutine with instrumentation is not supported. Any usage of tornado.gen.coroutine should be replaced with an asyncio coroutine.

Before

class MainHandler(tornado.web.RequestHandler):
    @tornado.gen.coroutine
    def get(self):

        yield gen.sleep(0.2)

After

class MainHandler(tornado.web.RequestHandler):
    @asyncio.coroutine
    def get(self):

        yield from gen.sleep(0.2)

Improved APIs for time trace context managers

Context managers (such as FunctionTrace, ExternalTrace, DatastoreTrace) may now be used from within asyncio coroutines! Additionally, the transaction argument is no longer used for any context manager API. In order to transition to the new API, the transaction argument must be removed.

Before

transaction = newrelic.agent.current_transaction()
with newrelic.agent.FunctionTrace(transaction, "my_trace"):
  pass

After

with newrelic.agent.FunctionTrace("my_trace"):
  pass

Updated WebTransaction context manager

The WebTransaction context manager is no longer WSGI dependent. The context manager takes generic HTTP arguments, which you can use to build web transactions that appear in the APM product. For more information, see our API documentation.

Notes

This release of the Python agent includes bug fixes, deprecations, and improved naming behavior for gRPC.

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.

Feature Enhancements

• Improved gRPC support

  The external and transaction pages now show the gRPC method being called in addition to the host and port.

• Ability to disable sending db.statement as part of traces

  When db.statement is added to either attributes.exclude or transaction_segments.attributes.exclude, the value will now no longer be sent to New Relic as part of transaction traces or spans.

Deprecations

• TransactionContext API is now deprecated

  The TransactionContext API will be removed in a future release.

• Tornado instrumentation + feature flag deprecation

  The existing introductory support for Tornado and feature flags will be removed from a future release. Versions 6.x and newer will be the only versions of Tornado supported in future releases. See this discussion on the New Relic forums for more details.

Bug Fixes

• Fixed a crash when using uvicorn workers with gunicorn

  When running ASGI applications with gunicorn, the agent would attempt to instrument the application as a WSGI application, resulting in a crash. The agent will no longer attempt to instrument gunicorn applications that are coroutines.

• Remove logging of license keys

  The agent logged license keys when data failed to send to New Relic. The agent will no longer log license keys to the agent logs under any circumstance.

• Fix operation of distributed tracing and cross application tracing when httplib connections are reused

  When making multiple requests via a single connection with httplib, httplib2, or urllib3, the proper headers will be added to each outgoing request. As a result, cross application tracing and distributed tracing will now operate as expected when reusing connection objects.

Notes

This release of the Python agent includes bug fixes and adds new API methods to retrieve the current trace ID and current span ID.

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

• Added a new API to retrieve the current trace ID

  Retrieving the current trace ID can now be accomplished with a call to newrelic.agent.current_trace_id(). This method will return the id of the current trace, or None if no transaction is in progress.

• Added a new API to retrieve the current span ID

  Retrieving the current span ID can now be accomplished with a call to newrelic.agent.current_span_id(). This method will return the id of the current span, or None if no span is in progress.

Bug Fixes

• Exclusive time incorrectly displayed when using concurrent asyncio

  The New Relic UI may have shown times equaling greater than 100 percent of the total time when executing concurrent tasks in asyncio. The exclusive times reported by the agent will now be correctly displayed in APM and insights.

Notes

This release of the Python agent adds improvements to the gRPC instrumentation package 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

• Improved gRPC support

  Support for measuring inbound gRPC requests has now been added. Inbound gRPC requests will now show as web transactions in APM. Additionally, distributed tracing is now supported as part of the gRPC instrumentation package.

Bug Fixes

• Update pika instrumentation to support the 1.x releases

  When using the agent with pika versions 1.x and up, the agent may have caused an application crash. The agent now correctly collects data on newer versions of pika.

• The startup_timeout setting may not have been honored in a multithreaded application

  When startup_timeout is set to a number greater than 0, the servicing of transactions is blocked until the agent has fully activated. In multithreaded applications, only the first thread would block due to improper locking. The agent will now block on all threads.

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.