sentry-sdk

New Features

• Additional integrations will now be activated automatically if the SDK detects the respective package is installed: Ariadne, ARQ, asyncpg, Chalice, clickhouse-driver, GQL, Graphene, huey, Loguru, PyMongo, Quart, Starlite, Strawberry.

Changed

• The Pyramid integration will not capture errors that might happen in authenticated_userid() in a custom AuthenticationPolicy class.

• The method need_code_loation of the MetricsAggregator was renamed to need_code_location.

• The BackgroundWorker thread used to process events was renamed from raven-sentry.BackgroundWorker to sentry-sdk.BackgroundWorker.

• The reraise function was moved from sentry_sdk._compat to sentry_sdk.utils.

• The _ScopeManager was moved from sentry_sdk.hub to sentry_sdk.scope.

• Moved the contents of tracing_utils_py3.py to tracing_utils.py. The start_child_span_decorator is now in sentry_sdk.tracing_utils.

• The actual implementation of get_current_span was moved to sentry_sdk.tracing_utils. sentry_sdk.get_current_span is still accessible as part of the top-level API.

• sentry_sdk.tracing_utils.get_current_span() does now take a scope instead of a hub as parameter.

• sentry_sdk.utils._get_contextvars does not return a tuple with three values, but a tuple with two values. The copy_context was removed.

• If you create a transaction manually and later mutate the transaction in a configure_scope block this does not work anymore. Here is a recipe on how to change your code to make it work: Your existing implementation:

  transaction = sentry_sdk.transaction(...)
  
  # later in the code execution:
  
  with sentry_sdk.configure_scope() as scope:
      scope.set_transaction_name("new-transaction-name")
  

  needs to be changed to this:

  transaction = sentry_sdk.transaction(...)
  
  # later in the code execution:
  
  scope = sentry_sdk.Scope.get_current_scope()
  scope.set_transaction_name("new-transaction-name")
  

• The classes listed in the table below are now abstract base classes. Therefore, they can no longer be instantiated. Subclasses can only be instantiated if they implement all of the abstract methods. <details> <summary><b>Show table</b></summary>

  Class	Abstract methods
  sentry_sdk.integrations.Integration	setup_once
  sentry_sdk.metrics.Metric	add, serialize_value, and weight
  sentry_sdk.profiler.Scheduler	setup and teardown
  sentry_sdk.transport.Transport	capture_envelope

  </details>

Removed

• Removed support for Python 2 and Python 3.5. The SDK now requires at least Python 3.6.
• Removed support for Celery 3.*.
• Removed support for Django 1.8, 1.9, 1.10.
• Removed support for Flask 0.*.
• Removed last_event_id() top level API. The last event ID is still returned by capture_event(), capture_exception() and capture_message() but the top level API sentry_sdk.last_event_id() has been removed.
• Removed support for sending events to the /store endpoint. Everything is now sent to the /envelope endpoint. If you're on SaaS you don't have to worry about this, but if you're running Sentry yourself you'll need version 20.6.0 or higher of self-hosted Sentry.
• The deprecated with_locals configuration option was removed. Use include_local_variables instead. See https://docs.sentry.io/platforms/python/configuration/options/#include-local-variables.
• The deprecated request_bodies configuration option was removed. Use max_request_body_size. See https://docs.sentry.io/platforms/python/configuration/options/#max-request-body-size.
• Removed support for user.segment. It was also removed from the trace header as well as from the dynamic sampling context.
• Removed support for the install method for custom integrations. Please use setup_once instead.
• Removed sentry_sdk.tracing.Span.new_span. Use sentry_sdk.tracing.Span.start_child instead.
• Removed sentry_sdk.tracing.Transaction.new_span. Use sentry_sdk.tracing.Transaction.start_child instead.
• Removed sentry_sdk.utils.Auth.store_api_url.
• sentry_sdk.utils.Auth.get_api_url's now accepts a sentry_sdk.consts.EndpointType enum instead of a string as its only parameter. We recommend omitting this argument when calling the function, since the parameter's default value is the only possible sentry_sdk.consts.EndpointType value. The parameter exists for future compatibility.
• Removed tracing_utils_py2.py. The start_child_span_decorator is now in sentry_sdk.tracing_utils.
• Removed the sentry_sdk.profiler.Scheduler.stop_profiling method. Any calls to this method can simply be removed, since this was a no-op method.

Deprecated

• profiler_mode and profiles_sample_rate have been deprecated as _experiments options. Use them as top level options instead:

  sentry_sdk.init(
      ...,
      profiler_mode="thread",
      profiles_sample_rate=1.0,
  )
  

• Deprecated sentry_sdk.transport.Transport.capture_event. Please use sentry_sdk.transport.Transport.capture_envelope, instead.
• Passing a function to sentry_sdk.init's transport keyword argument has been deprecated. If you wish to provide a custom transport, please pass a sentry_sdk.transport.Transport instance or a subclass.
• The parameter propagate_hub in ThreadingIntegration() was deprecated and renamed to propagate_scope.

Sentry SDK 2.0.0a2 is alpha software and not yet ready for production!

Please give it a spin and test it with your project. If you have any questions or feedback please contact us on Discord in the #python channel or create a GitHub Issue or start a GitHub Discussion.

Thanks!

New Features

• Additional integrations will now be activated automatically if the SDK detects the respective package is installed: Ariadne, ARQ, asyncpg, Chalice, clickhouse-driver, GQL, Graphene, huey, Loguru, PyMongo, Quart, Starlite, Strawberry.

Changed

• The Pyramid integration will not capture errors that might happen in authenticated_userid() in a custom AuthenticationPolicy class.

• The method need_code_loation of the MetricsAggregator was renamed to need_code_location.

• The BackgroundWorker thread used to process events was renamed from raven-sentry.BackgroundWorker to sentry-sdk.BackgroundWorker.

• The reraise function was moved from sentry_sdk._compat to sentry_sdk.utils.

• The _ScopeManager was moved from sentry_sdk.hub to sentry_sdk.scope.

• Moved the contents of tracing_utils_py3.py to tracing_utils.py. The start_child_span_decorator is now in sentry_sdk.tracing_utils.

• The actual implementation of get_current_span was moved to sentry_sdk.tracing_utils. sentry_sdk.get_current_span is still accessible as part of the top-level API.

• sentry_sdk.tracing_utils.get_current_span() does now take a scope instead of a hub as parameter.

• sentry_sdk.utils._get_contextvars does not return a tuple with three values, but a tuple with two values. The copy_context was removed.

• If you create a transaction manually and later mutate the transaction in a configure_scope block this does not work anymore. Here is a recipe on how to change your code to make it work: Your existing implementation:

  transaction = sentry_sdk.transaction(...)
  
  # later in the code execution:
  
  with sentry_sdk.configure_scope() as scope:
      scope.set_transaction_name("new-transaction-name")
  

  needs to be changed to this:

  transaction = sentry_sdk.transaction(...)
  
  # later in the code execution:
  
  scope = sentry_sdk.Scope.get_current_scope()
  scope.set_transaction_name("new-transaction-name")
  

• The classes listed in the table below are now abstract base classes. Therefore, they can no longer be instantiated. Subclasses can only be instantiated if they implement all of the abstract methods. <details> <summary><b>Show table</b></summary>

  Class	Abstract methods
  sentry_sdk.integrations.Integration	setup_once
  sentry_sdk.metrics.Metric	add, serialize_value, and weight
  sentry_sdk.profiler.Scheduler	setup and teardown
  sentry_sdk.transport.Transport	capture_envelope

  </details>

Removed

• Removed support for Python 2 and Python 3.5. The SDK now requires at least Python 3.6.
• Removed support for Celery 3.*.
• Removed support for Django 1.8, 1.9, 1.10.
• Removed support for Flask 0.*.
• Removed last_event_id() top level API. The last event ID is still returned by capture_event(), capture_exception() and capture_message() but the top level API sentry_sdk.last_event_id() has been removed.
• Removed support for sending events to the /store endpoint. Everything is now sent to the /envelope endpoint. If you're on SaaS you don't have to worry about this, but if you're running Sentry yourself you'll need version 20.6.0 or higher of self-hosted Sentry.
• The deprecated with_locals configuration option was removed. Use include_local_variables instead. See https://docs.sentry.io/platforms/python/configuration/options/#include-local-variables.
• The deprecated request_bodies configuration option was removed. Use max_request_body_size. See https://docs.sentry.io/platforms/python/configuration/options/#max-request-body-size.
• Removed support for user.segment. It was also removed from the trace header as well as from the dynamic sampling context.
• Removed support for the install method for custom integrations. Please use setup_once instead.
• Removed sentry_sdk.tracing.Span.new_span. Use sentry_sdk.tracing.Span.start_child instead.
• Removed sentry_sdk.tracing.Transaction.new_span. Use sentry_sdk.tracing.Transaction.start_child instead.
• Removed sentry_sdk.utils.Auth.store_api_url.
• sentry_sdk.utils.Auth.get_api_url's now accepts a sentry_sdk.consts.EndpointType enum instead of a string as its only parameter. We recommend omitting this argument when calling the function, since the parameter's default value is the only possible sentry_sdk.consts.EndpointType value. The parameter exists for future compatibility.
• Removed tracing_utils_py2.py. The start_child_span_decorator is now in sentry_sdk.tracing_utils.
• Removed the sentry_sdk.profiler.Scheduler.stop_profiling method. Any calls to this method can simply be removed, since this was a no-op method.

Deprecated

• profiler_mode and profiles_sample_rate have been deprecated as _experiments options. Use them as top level options instead:

  sentry_sdk.init(
      ...,
      profiler_mode="thread",
      profiles_sample_rate=1.0,
  )
  

• Deprecated sentry_sdk.transport.Transport.capture_event. Please use sentry_sdk.transport.Transport.capture_envelope, instead.
• Passing a function to sentry_sdk.init's transport keyword argument has been deprecated. If you wish to provide a custom transport, please pass a sentry_sdk.transport.Transport instance or a subclass.
• The parameter propagate_hub in ThreadingIntegration() was deprecated and renamed to propagate_scope.

Sentry SDK 2.0a1 is alpha software and not yet ready for production!

Please give it a spin and test it with your project. If you have any questions or feedback please contact us on Discord in the #python channel or create a GitHub Issue or start a GitHub Discussion.

Thanks!

New Features

• Additional integrations will now be activated automatically if the SDK detects the respective package is installed: Ariadne, ARQ, asyncpg, Chalice, clickhouse-driver, GQL, Graphene, huey, Loguru, PyMongo, Quart, Starlite, Strawberry.

Changed

• The Pyramid integration will not capture errors that might happen in authenticated_userid() in a custom AuthenticationPolicy class.

• The method need_code_loation of the MetricsAggregator was renamed to need_code_location.

• The BackgroundWorker thread used to process events was renamed from raven-sentry.BackgroundWorker to sentry-sdk.BackgroundWorker.

• The reraise function was moved from sentry_sdk._compat to sentry_sdk.utils.

• The _ScopeManager was moved from sentry_sdk.hub to sentry_sdk.scope.

• Moved the contents of tracing_utils_py3.py to tracing_utils.py. The start_child_span_decorator is now in sentry_sdk.tracing_utils.

• The actual implementation of get_current_span was moved to sentry_sdk.tracing_utils. sentry_sdk.get_current_span is still accessible as part of the top-level API.

• sentry_sdk.tracing_utils.get_current_span() does now take a scope instead of a hub as parameter.

• sentry_sdk.utils._get_contextvars does not return a tuple with three values, but a tuple with two values. The copy_context was removed.

• If you create a transaction manually and later mutate the transaction in a configure_scope block this does not work anymore. Here is a recipe on how to change your code to make it work: Your existing implementation:

  transaction = sentry_sdk.transaction(...)
  
  # later in the code execution:
  
  with sentry_sdk.configure_scope() as scope:
      scope.set_transaction_name("new-transaction-name")
  

  needs to be changed to this:

  transaction = sentry_sdk.transaction(...)
  
  # later in the code execution:
  
  scope = sentry_sdk.Scope.get_current_scope()
  scope.set_transaction_name("new-transaction-name")
  

• The classes listed in the table below are now abstract base classes. Therefore, they can no longer be instantiated. Subclasses can only be instantiated if they implement all of the abstract methods. <details> <summary><b>Show table</b></summary>

  Class	Abstract methods
  sentry_sdk.integrations.Integration	setup_once
  sentry_sdk.metrics.Metric	add, serialize_value, and weight
  sentry_sdk.profiler.Scheduler	setup and teardown
  sentry_sdk.transport.Transport	capture_envelope

  </details>

Removed

• Removed support for Python 2 and Python 3.5. The SDK now requires at least Python 3.6.
• Removed support for Celery 3.*.
• Removed support for Django 1.8, 1.9, 1.10.
• Removed support for Flask 0.*.
• Removed last_event_id() top level API. The last event ID is still returned by capture_event(), capture_exception() and capture_message() but the top level API sentry_sdk.last_event_id() has been removed.
• Removed support for sending events to the /store endpoint. Everything is now sent to the /envelope endpoint. If you're on SaaS you don't have to worry about this, but if you're running Sentry yourself you'll need version 20.6.0 or higher of self-hosted Sentry.
• The deprecated with_locals configuration option was removed. Use include_local_variables instead. See https://docs.sentry.io/platforms/python/configuration/options/#include-local-variables.
• The deprecated request_bodies configuration option was removed. Use max_request_body_size. See https://docs.sentry.io/platforms/python/configuration/options/#max-request-body-size.
• Removed support for user.segment. It was also removed from the trace header as well as from the dynamic sampling context.
• Removed support for the install method for custom integrations. Please use setup_once instead.
• Removed sentry_sdk.tracing.Span.new_span. Use sentry_sdk.tracing.Span.start_child instead.
• Removed sentry_sdk.tracing.Transaction.new_span. Use sentry_sdk.tracing.Transaction.start_child instead.
• Removed sentry_sdk.utils.Auth.store_api_url.
• sentry_sdk.utils.Auth.get_api_url's now accepts a sentry_sdk.consts.EndpointType enum instead of a string as its only parameter. We recommend omitting this argument when calling the function, since the parameter's default value is the only possible sentry_sdk.consts.EndpointType value. The parameter exists for future compatibility.
• Removed tracing_utils_py2.py. The start_child_span_decorator is now in sentry_sdk.tracing_utils.
• Removed the sentry_sdk.profiler.Scheduler.stop_profiling method. Any calls to this method can simply be removed, since this was a no-op method.

Deprecated

• profiler_mode and profiles_sample_rate have been deprecated as _experiments options. Use them as top level options instead:

  sentry_sdk.init(
      ...,
      profiler_mode="thread",
      profiles_sample_rate=1.0,
  )
  

• Deprecated sentry_sdk.transport.Transport.capture_event. Please use sentry_sdk.transport.Transport.capture_envelope, instead.
• Passing a function to sentry_sdk.init's transport keyword argument has been deprecated. If you wish to provide a custom transport, please pass a sentry_sdk.transport.Transport instance or a subclass.
• The parameter propagate_hub in ThreadingIntegration() was deprecated and renamed to propagate_scope.

This is a security backport release.

• Don't send full env to subprocess (892dd800) by @kmichel-aiven

  See also https://github.com/getsentry/sentry-python/security/advisories/GHSA-g92j-qhmh-64v2

This is the final 1.x release for the forseeable future. Development will continue on the 2.x release line. The first 2.x version will be available in the next few weeks.

Various fixes & improvements

• Allow to upsert monitors (#2929) by @sentrivana

  It's now possible to provide monitor_config to the monitor decorator/context manager directly:

  from sentry_sdk.crons import monitor
  
  # All keys except `schedule` are optional
  monitor_config = {
      "schedule": {"type": "crontab", "value": "0 0 * * *"},
      "timezone": "Europe/Vienna",
      "checkin_margin": 10,
      "max_runtime": 10,
      "failure_issue_threshold": 5,
      "recovery_threshold": 5,
  }
  
  @monitor(monitor_slug='<monitor-slug>', monitor_config=monitor_config)
  def tell_the_world():
      print('My scheduled task...')
  

  Check out the cron docs for details.

• Add Django signals_denylist to filter signals that are attached to by signals_spans (#2758) by @lieryan

  If you want to exclude some Django signals from performance tracking, you can use the new signals_denylist Django option:

  import django.db.models.signals
  import sentry_sdk
  
  sentry_sdk.init(
      ...
      integrations=[
          DjangoIntegration(
              ...
              signals_denylist=[
                  django.db.models.signals.pre_init, 
                  django.db.models.signals.post_init,
              ],
          ),
      ],
  )
  

• increment for metrics (#2588) by @mitsuhiko

  increment and incr are equivalent, so you can pick whichever you like more.

• Add value, unit to before_emit_metric (#2958) by @sentrivana

  If you add a custom before_emit_metric, it'll now accept 4 arguments (the key, value, unit and tags) instead of just key and tags.

  def before_emit(key, value, unit, tags):
      if key == "removed-metric":
          return False
      tags["extra"] = "foo"
      del tags["release"]
      return True
  
  sentry_sdk.init(
      ...
      _experiments={
          "before_emit_metric": before_emit,
      }
  )
  

• Remove experimental metric summary options (#2957) by @sentrivana

  The _experiments options metrics_summary_sample_rate and should_summarize_metric have been removed.

• New normalization rules for metric keys, names, units, tags (#2946) by @sentrivana

• Change data_category from statsd to metric_bucket (#2954) by @cleptric

• Accessing __mro__ might throw a ValueError (#2952) by @sentrivana

• Suppress prompt spawned by subprocess when using pythonw (#2936) by @collinbanko

• Handle None in GraphQL query #2715 (#2762) by @czyber

• Do not send "quiet" Sanic exceptions to Sentry (#2821) by @hamedsh

• Implement metric_bucket rate limits (#2933) by @cleptric

• Fix type hints for monitor decorator (#2944) by @szokeasaurusrex

• Remove deprecated typing imports in crons (#2945) by @szokeasaurusrex

• Make monitor_config a TypedDict (#2931) by @sentrivana

• Add devenv-requirements.txt and update env setup instructions (#2761) by @arr-ee

• Bump types-protobuf from 4.24.0.20240311 to 4.24.0.20240408 (#2941) by @dependabot

• Disable Codecov check run annotations (#2537) by @eliatcodecov

Various fixes & improvements

• Make monitor async friendly (#2912) by @sentrivana

  You can now decorate your async functions with the monitor decorator and they will correctly report their duration and completion status.

• Fixed Event | None runtime TypeError (#2928) by @szokeasaurusrex

Various fixes & improvements

• ref: Define types at runtime (#2914) by @szokeasaurusrex
• Explicit reexport of types (#2866) (#2913) by @szokeasaurusrex
• feat(profiling): Add thread data to spans (#2843) by @Zylphrex

Various fixes & improvements

• Add optional keep_alive (#2842) by @sentrivana

  If you're experiencing frequent network issues between the SDK and Sentry, you can try turning on TCP keep-alive:

  import sentry_sdk
  
  sentry_sdk.init(
      # ...your usual settings...
      keep_alive=True,
  )
  

• Add support for Celery Redbeat cron tasks (#2643) by @kwigley

  The SDK now supports the Redbeat scheduler in addition to the default Celery Beat scheduler for auto instrumenting crons. See the docs for more information about how to set this up.

• aws_event can be an empty list (#2849) by @sentrivana

• Re-export Event in types.py (#2829) by @szokeasaurusrex

• Small API docs improvement (#2828) by @antonpirker

• Fixed OpenAI tests (#2834) by @antonpirker

• Bump checkouts/data-schemas from ed078ed to 8232f17 (#2832) by @dependabot

Various fixes & improvements

• New integration: OpenAI integration (#2791) by @colin-sentry

  We added an integration for OpenAI to capture errors and also performance data when using the OpenAI Python SDK.

  Useage:

  This integrations is auto-enabling, so if you have the openai package in your project it will be enabled. Just initialize Sentry before you create your OpenAI client.

  from openai import OpenAI
  
  import sentry_sdk
  
  sentry_sdk.init(
      dsn="___PUBLIC_DSN___",
      enable_tracing=True,
      traces_sample_rate=1.0,
  )
  
  client = OpenAI()
  

  For more information, see the documentation for OpenAI integration.

• Discard open OpenTelemetry spans after 10 minutes (#2801) by @antonpirker

• Propagate sentry-trace and baggage headers to Huey tasks (#2792) by @cnschn

• Added Event type (#2753) by @szokeasaurusrex

• Improve scrub_dict typing (#2768) by @szokeasaurusrex

• Dependencies: bump types-protobuf from 4.24.0.20240302 to 4.24.0.20240311 (#2797) by @dependabot

Various fixes & improvements

• Add recursive scrubbing to EventScrubber (#2755) by @Cheapshot003

  By default, the EventScrubber will not search your events for potential PII recursively. With this release, you can enable this behavior with:

  import sentry_sdk
  from sentry_sdk.scrubber import EventScrubber
  
  sentry_sdk.init(
      # ...your usual settings...
      event_scrubber=EventScrubber(recursive=True),
  )
  

• Expose socket_options (#2786) by @sentrivana

  If the SDK is experiencing connection issues (connection resets, server closing connection without response, etc.) while sending events to Sentry, tweaking the default urllib3 socket options to the following can help:

  import socket
  from urllib3.connection import HTTPConnection
  import sentry_sdk
  
  sentry_sdk.init(
      # ...your usual settings...
      socket_options=HTTPConnection.default_socket_options + [
          (socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1),
          # note: skip the following line if you're on MacOS since TCP_KEEPIDLE doesn't exist there
          (socket.SOL_TCP, socket.TCP_KEEPIDLE, 45),
          (socket.SOL_TCP, socket.TCP_KEEPINTVL, 10),
          (socket.SOL_TCP, socket.TCP_KEEPCNT, 6),
      ],
  )
  

• Allow to configure merge target for releases (#2777) by @sentrivana

• Allow empty character in metric tags values (#2775) by @viglia

• Replace invalid tag values with an empty string instead of _ (#2773) by @markushi

• Add documentation comment to scrub_list (#2769) by @szokeasaurusrex

• Fixed regex to parse version in lambda package file (#2767) by @antonpirker

• xfail broken AWS Lambda tests for now (#2794) by @sentrivana

• Removed print statements because it messes with the tests (#2789) by @antonpirker

• Bump types-protobuf from 4.24.0.20240129 to 4.24.0.20240302 (#2782) by @dependabot

• Bump checkouts/data-schemas from eb941c2 to ed078ed (#2781) by @dependabot

Various fixes & improvements

• Fix compatibility with greenlet/gevent (#2756) by @sentrivana
• Fix query source relative filepath (#2717) by @gggritso
• Support clickhouse-driver==0.2.7 (#2752) by @sentrivana
• Bump checkouts/data-schemas from 6121fd3 to eb941c2 (#2747) by @dependabot

Various fixes & improvements

• Deprecate last_event_id(). (#2749) by @antonpirker

• Warn if uWSGI is set up without proper thread support (#2738) by @sentrivana

  uWSGI has to be run in threaded mode for the SDK to run properly. If this is not the case, the consequences could range from features not working unexpectedly to uWSGI workers crashing.

  Please make sure to run uWSGI with both --enable-threads and --py-call-uwsgi-fork-hooks.

• parsed_url can be None (#2734) by @sentrivana

• Python 3.7 is not supported anymore by Lambda, so removed it and added 3.12 (#2729) by @antonpirker

Various fixes & improvements

• Only start metrics flusher thread on demand (#2727) by @sentrivana
• Bump checkouts/data-schemas from aa7058c to 6121fd3 (#2724) by @dependabot

Various fixes & improvements

• Turn off metrics for uWSGI (#2720) by @sentrivana
• Minor improvements (#2714) by @antonpirker

Various fixes & improvements

• test: Fix pytest error (#2712) by @szokeasaurusrex
• build(deps): bump types-protobuf from 4.24.0.4 to 4.24.0.20240129 (#2691) by @dependabot

Various fixes & improvements

• Fix uWSGI workers hanging (#2694) by @sentrivana
• Make metrics work with gevent (#2694) by @sentrivana
• Guard against engine.url being None (#2708) by @sentrivana
• Fix performance regression in sentry_sdk.utils._generate_installed_modules (#2703) by @GlenWalker
• Guard against Sentry initialization mid SQLAlchemy cursor (#2702) by @apmorton
• Fix yaml generation script (#2695) by @sentrivana
• Fix AWS Lambda workflow (#2710) by @sentrivana
• Bump codecov/codecov-action from 3 to 4 (#2706) by @dependabot
• Bump actions/cache from 3 to 4 (#2661) by @dependabot
• Bump actions/checkout from 3.1.0 to 4.1.1 (#2561) by @dependabot
• Bump github/codeql-action from 2 to 3 (#2603) by @dependabot
• Bump actions/setup-python from 4 to 5 (#2577) by @dependabot

Various fixes & improvements

• Enable metrics related settings by default (#2685) by @iambriccardo
• Fix UnicodeDecodeError on Python 2 (#2657) by @sentrivana
• Enable DB query source by default (#2629) by @sentrivana
• Fix query source duration check (#2675) by @sentrivana
• Reformat with black==24.1.0 (#2680) by @sentrivana
• Cleaning up existing code to prepare for new Scopes API (#2611) by @antonpirker
• Moved redis related tests to databases (#2674) by @antonpirker
• Improve sentry_sdk.trace type hints (#2633) by @szokeasaurusrex
• Bump checkouts/data-schemas from e9f7d58 to aa7058c (#2639) by @dependabot

Various fixes & improvements

• Fix timestamp in transaction created by OTel (#2627) by @antonpirker
• Fix relative path in DB query source (#2624) by @antonpirker
• Run more CI checks on 2.0 branch (#2625) by @sentrivana
• Fix tracing TypeError for static and class methods (#2559) by @szokeasaurusrex
• Fix missing ctx in Arq integration (#2600) by @ivanovart
• Change data_category from check_in to monitor (#2598) by @sentrivana

Various fixes & improvements

• Fix psycopg2 detection in the Django integration (#2593) by @sentrivana
• Filter out empty string releases (#2591) by @sentrivana
• Fixed local var not present when there is an error in a user's error_sampler function (#2511) by @antonpirker
• Fixed typing in aiohttp (#2590) by @antonpirker

Various fixes & improvements

• Add support for cluster clients from Redis SDK (#2394) by @md384
• Improve location reporting for timer metrics (#2552) by @mitsuhiko
• Fix Celery TypeError with no-argument apply_async (#2575) by @szokeasaurusrex
• Fix Lambda integration with EventBridge source (#2546) by @davidcroda
• Add max tries to Spotlight (#2571) by @hazAT
• Handle os.path.devnull access issues (#2579) by @sentrivana
• Change code.filepath frame picking logic (#2568) by @sentrivana
• Trigger AWS Lambda tests on label (#2538) by @sentrivana
• Run permissions step on pull_request_target but not push (#2548) by @sentrivana
• Hash AWS Lambda test functions based on current revision (#2557) by @sentrivana
• Update Django version in tests (#2562) by @sentrivana
• Make metrics tests non-flaky (#2572) by @antonpirker

Various fixes & improvements

• Only add trace context to checkins and do not run event_processors for checkins (#2536) by @antonpirker
• Metric span summaries (#2522) by @mitsuhiko
• Add source context to code locations (#2539) by @jan-auer
• Use in-app filepath instead of absolute path (#2541) by @antonpirker
• Switch to jinja2 for generating CI yamls (#2534) by @sentrivana

Various fixes & improvements

• Fix NameError on parse_version with eventlet (#2532) by @sentrivana
• build(deps): bump checkouts/data-schemas from 68def1e to e9f7d58 (#2501) by @dependabot

Various fixes & improvements

• Move installed modules code to utils (#2429) by @sentrivana

  Note: We moved the internal function _get_installed_modules from sentry_sdk.integrations.modules to sentry_sdk.utils. So if you use this function you have to update your imports

• Add code locations for metrics (#2526) by @jan-auer

• Add query source to DB spans (#2521) by @antonpirker

• Send events to Spotlight sidecar (#2524) by @HazAT

• Run integration tests with newest pytest (#2518) by @sentrivana

• Bring tests up to date (#2512) by @sentrivana

• Fix: Prevent global var from being discarded at shutdown (#2530) by @antonpirker

• Fix: Scope transaction source not being updated in scope.span setter (#2519) by @sl0thentr0py

• Django: Support Django 5.0 (#2490) by @sentrivana
• Django: Handling ASGI body in the right way. (#2513) by @antonpirker
• Flask: Test with Flask 3.0 (#2506) by @sentrivana
• Celery: Do not create a span when task is triggered by Celery Beat (#2510) by @antonpirker
• Redis: Ensure RedisIntegration is disabled, unless redis is installed (#2504) by @szokeasaurusrex
• Quart: Fix Quart integration for Quart 0.19.4 (#2516) by @antonpirker
• gRPC: Make async gRPC less noisy (#2507) by @jyggen

Various fixes & improvements

• Updated gRPC integration: Asyncio interceptors and easier setup (#2369) by @fdellekart

  Our gRPC integration now instruments incoming unary-unary grpc requests and outgoing unary-unary, unary-stream grpc requests using grpcio channels. Everything works now for sync and async code.

  Before this release you had to add Sentry interceptors by hand to your gRPC code, now the only thing you need to do is adding the GRPCIntegration to you sentry_sdk_init() call. (See documentation for more information):

  import sentry_sdk
  from sentry_sdk.integrations.grpc import GRPCIntegration
  
  sentry_sdk.init(
      dsn="___PUBLIC_DSN___",
      enable_tracing=True,
      integrations=[
          GRPCIntegration(),
      ],
  )
  

  The old way still works, but we strongly encourage you to update your code to the way described above.

• Python 3.12: Replace deprecated datetime functions (#2502) by @sentrivana

• Metrics: Unify datetime format (#2409) by @mitsuhiko

• Celery: Set correct data in check_ins (#2500) by @antonpirker

• Celery: Read timezone for Crons monitors from celery_schedule if existing (#2497) by @antonpirker

• Django: Removing redundant code in Django tests (#2491) by @vagi8

• Django: Make reading the request body work in Django ASGI apps. (#2495) by @antonpirker

• FastAPI: Use wraps on fastapi request call wrapper (#2476) by @nkaras

• Fix: Probe for psycopg2 and psycopg3 parameters function. (#2492) by @antonpirker

• Fix: Remove unnecessary TYPE_CHECKING alias (#2467) by @rafrafek

Various fixes & improvements

• Added Python 3.12 support (#2471, #2483)
• Handle missing connection_kwargs in patch_redis_client (#2482) by @szokeasaurusrex
• Run common test suite on Python 3.12 (#2479) by @sentrivana

Various fixes & improvements

• Make parse_version work in utils.py itself. (#2474) by @antonpirker

Various fixes & improvements

• New: Added error_sampler option (#2456) by @szokeasaurusrex
• Python 3.12: Detect interpreter in shutdown state on thread spawn (#2468) by @mitsuhiko
• Patch eventlet under Sentry SDK (#2464) by @szokeasaurusrex
• Mitigate CPU spikes when sending lots of events with lots of data (#2449) by @antonpirker
• Make debug option also configurable via environment (#2450) by @antonpirker
• Make sure get_dsn_parameters is an actual function (#2441) by @sentrivana
• Bump pytest-localserver, add compat comment (#2448) by @sentrivana
• AWS Lambda: Update compatible runtimes for AWS Lambda layer (#2453) by @antonpirker
• AWS Lambda: Load AWS Lambda secrets in Github CI (#2153) by @antonpirker
• Redis: Connection attributes in redis database spans (#2398) by @antonpirker
• Falcon: Falcon integration checks response status before reporting error (#2465) by @szokeasaurusrex
• Quart: Support Quart 0.19 onwards (#2403) by @pgjones
• Sanic: Sanic integration initial version (#2419) by @szokeasaurusrex
• Django: Fix parsing of Django path patterns (#2452) by @sentrivana
• Django: Add Django 4.2 to test suite (#2462) by @sentrivana
• Polish changelog (#2434) by @sentrivana
• Update CONTRIBUTING.md (#2443) by @krishvsoni
• Update README.md (#2435) by @sentrivana

Various fixes & improvements

• New: Error monitoring for some of the most popular Python GraphQL libraries:

  • Add GQL GraphQL integration (#2368) by @szokeasaurusrex

    Usage:

      import sentry_sdk
      from sentry_sdk.integrations.gql import GQLIntegration
    
      sentry_sdk.init(
          dsn='___PUBLIC_DSN___',
          integrations=[
              GQLIntegration(),
          ],
      )
    

  • Add Graphene GraphQL error integration (#2389) by @sentrivana

    Usage:

      import sentry_sdk
      from sentry_sdk.integrations.graphene import GrapheneIntegration
    
      sentry_sdk.init(
          dsn='___PUBLIC_DSN___',
          integrations=[
              GrapheneIntegration(),
          ],
      )
    

  • Add Strawberry GraphQL error & tracing integration (#2393) by @sentrivana

    Usage:

      import sentry_sdk
      from sentry_sdk.integrations.strawberry import StrawberryIntegration
    
      sentry_sdk.init(
          dsn='___PUBLIC_DSN___',
          integrations=[
              # make sure to set async_execution to False if you're executing
              # GraphQL queries synchronously
              StrawberryIntegration(async_execution=True),
          ],
          traces_sample_rate=1.0,
      )
    

  • Add Ariadne GraphQL error integration (#2387) by @sentrivana

    Usage:

      import sentry_sdk
      from sentry_sdk.integrations.ariadne import AriadneIntegration
    
      sentry_sdk.init(
          dsn='___PUBLIC_DSN___',
          integrations=[
              AriadneIntegration(),
          ],
      )
    

• Capture multiple named groups again (#2432) by @sentrivana

• Don't fail when upstream scheme is unusual (#2371) by @vanschelven

• Support new RQ version (#2405) by @antonpirker

• Remove utcnow, utcfromtimestamp deprecated in Python 3.12 (#2415) by @rmad17

• Add trace to __all__ in top-level __init__.py (#2401) by @lobsterkatie

• Move minimetrics code to the SDK (#2385) by @mitsuhiko

• Add configurable compression levels (#2382) by @mitsuhiko

• Shift flushing by up to a rollup window (#2396) by @mitsuhiko

• Make a consistent noop flush behavior (#2428) by @mitsuhiko

• Stronger recursion protection (#2426) by @mitsuhiko

• Remove OpenTelemetryIntegration from __init__.py (#2379) by @sentrivana

• Update API docs (#2397) by @antonpirker

• Pin some test requirements because new majors break our tests (#2404) by @antonpirker

• Run more requests, celery, falcon tests (#2414) by @sentrivana

• Move importorskips in tests to __init__.py files (#2412) by @sentrivana

• Fix mypy errors (#2433) by @sentrivana

• Fix pre-commit issues (#2424) by @bukzor-sentryio

• Update CONTRIBUTING.md (#2411) by @sentrivana

• Bump sphinx from 7.2.5 to 7.2.6 (#2378) by @dependabot

• [Experimental] Add explain plan to DB spans (#2315) by @antonpirker

Various fixes & improvements

• New: Add integration for clickhouse-driver (#2167) by @mimre25

  For more information, see the documentation for clickhouse-driver for more information.

  Usage:

    import sentry_sdk
    from sentry_sdk.integrations.clickhouse_driver import ClickhouseDriverIntegration
  
    sentry_sdk.init(
        dsn='___PUBLIC_DSN___',
        integrations=[
            ClickhouseDriverIntegration(),
        ],
    )
  

• New: Add integration for asyncpg (#2314) by @mimre25

  For more information, see the documentation for asyncpg for more information.

  Usage:

    import sentry_sdk
    from sentry_sdk.integrations.asyncpg import AsyncPGIntegration
  
    sentry_sdk.init(
        dsn='___PUBLIC_DSN___',
        integrations=[
            AsyncPGIntegration(),
        ],
    )
  

• New: Allow to override propagate_traces in Celery per task (#2331) by @jan-auer

  For more information, see the documentation for Celery for more information.

  Usage:

    import sentry_sdk
    from sentry_sdk.integrations.celery import CeleryIntegration
  
    # Enable global distributed traces (this is the default, just to be explicit.)
    sentry_sdk.init(
        dsn='___PUBLIC_DSN___',
        integrations=[
            CeleryIntegration(propagate_traces=True),
        ],
    )
  
    ...
  
    # This will NOT propagate the trace. (The task will start its own trace):
    my_task_b.apply_async(
        args=("some_parameter", ),
        headers={"sentry-propagate-traces": False},
    )
  

• Prevent Falcon integration from breaking ASGI apps (#2359) by @szokeasaurusrex

• Backpressure: only downsample a max of 10 times (#2347) by @sl0thentr0py

• Made NoOpSpan compatible to Transactions. (#2364) by @antonpirker

• Cleanup ASGI integration (#2335) by @antonpirker

• Pin anyio in tests (dep of httpx), because new major 4.0.0 breaks tests. (#2336) by @antonpirker

• Added link to backpressure section in docs. (#2354) by @antonpirker

• Add .vscode to .gitignore (#2317) by @shoaib-mohd

• Documenting Spans and Transactions (#2358) by @antonpirker

• Fix in profiler: do not call getcwd from module root (#2329) by @Zylphrex

• Fix deprecated version attribute (#2338) by @vagi8

• Fix transaction name in Starlette and FastAPI (#2341) by @antonpirker

• Fix tests using Postgres (#2362) by @antonpirker

• build(deps): Updated linting tooling (#2350) by @antonpirker

• build(deps): bump sphinx from 7.2.4 to 7.2.5 (#2344) by @dependabot

• build(deps): bump actions/checkout from 2 to 4 (#2352) by @dependabot

• build(deps): bump checkouts/data-schemas from ebc77d3 to 68def1e (#2351) by @dependabot

Various fixes & improvements

• Officially support Python 3.11 (#2300) by @sentrivana

• Context manager monitor (#2290) by @szokeasaurusrex

• Set response status code in transaction response context. (#2312) by @antonpirker

• Add missing context kwarg to _sentry_task_factory (#2267) by @JohnnyDeuss

• In Postgres take the connection params from the connection (#2308) by @antonpirker

• Experimental: Allow using OTel for performance instrumentation (#2272) by @sentrivana

  This release includes experimental support for replacing Sentry's default performance monitoring solution with one powered by OpenTelemetry without having to do any manual setup.

  Try it out by installing pip install sentry-sdk[opentelemetry-experimental] and then initializing the SDK with:

  sentry_sdk.init(
      # ...your usual options...
      _experiments={"otel_powered_performance": True},
  )
  

  This enables OpenTelemetry performance monitoring support for some of the most popular frameworks and libraries (Flask, Django, FastAPI, requests...).

  We're looking forward to your feedback! Please let us know about your experience in this discussion: https://github.com/getsentry/sentry/discussions/55023

  Important note: Please note that this feature is experimental and in a proof-of-concept stage and is not meant for production use. It may be changed or removed at any point.

• Enable backpressure handling by default (#2298) by @sl0thentr0py

  The SDK now dynamically downsamples transactions to reduce backpressure in high throughput systems. It starts a new Monitor thread to perform some health checks which decide to downsample (halved each time) in 10 second intervals till the system is healthy again.

  To disable this behavior, use:

  sentry_sdk.init(
      # ...your usual options...
      enable_backpressure_handling=False,
  )
  

  If your system serves heavy load, please let us know how this feature works for you! Check out the documentation for more information.

• Stop recording spans for internal web requests to Sentry (#2297) by @szokeasaurusrex

• Add test for ThreadPoolExecutor (#2259) by @gggritso

• Add docstrings for Scope.update_from_* (#2311) by @sentrivana

• Moved is_sentry_url to utils (#2304) by @szokeasaurusrex

• Fix: arq attribute error on settings, support worker args (#2260) by @rossmacarthur

• Fix: Exceptions include detail property for their value (#2193) by @nicolassanmar

• build(deps): bump mypy from 1.4.1 to 1.5.1 (#2319) by @dependabot

• build(deps): bump sphinx from 7.1.2 to 7.2.4 (#2322) by @dependabot

• build(deps): bump sphinx from 7.0.1 to 7.1.2 (#2296) by @dependabot

• build(deps): bump checkouts/data-schemas from 1b85152 to ebc77d3 (#2254) by @dependabot

Various fixes & improvements

• Revert GraphQL integration (#2287) by @sentrivana

Various fixes & improvements

• Fix GraphQL integration swallowing responses (#2286) by @sentrivana
• Fix typo (#2283) by @sentrivana

Various fixes & improvements

• Capture GraphQL client errors (#2243) by @sentrivana
  • The SDK will now create dedicated errors whenever an HTTP client makes a reqwuest to a /graphql endpoint and the response contains an error. You can opt out of this by providing capture_graphql_errors=False to the HTTP client integration.
• Read MAX_VALUE_LENGTH from client options (#2121) (#2171) by @puittenbroek
• Rename request_bodies to max_request_body_size (#2247) by @mgaligniana
• Always sample checkin regardless of sample_rate (#2279) by @szokeasaurusrex
• Add information to short-interval cron error message (#2246) by @lobsterkatie
• Add DB connection attributes in spans (#2274) by @antonpirker
• Add db.system to remaining Redis spans (#2271) by @AbhiPrasad
• Clarified the procedure for running tests (#2276) by @szokeasaurusrex
• Fix Chalice tests (#2278) by @sentrivana
• Bump Black from 23.3.0 to 23.7.0 (#2256) by @dependabot
• Remove py3.4 from tox.ini (#2248) by @sentrivana

Various fixes & improvements

• Redis: Add support for redis.asyncio (#1933) by @Zhenay
• Make sure each task that is started by Celery Beat has its own trace. (#2249) by @antonpirker
• Add Sampling Decision to Trace Envelope Header (#2239) by @antonpirker
• Do not add trace headers (sentry-trace and baggage) to HTTP requests to Sentry (#2240) by @antonpirker
• Prevent adding sentry-trace header multiple times (#2235) by @antonpirker
• Skip distributions with incomplete metadata (#2231) by @rominf
• Remove stale.yml (#2245) by @hubertdeng123
• Django: Fix 404 Handler handler being labeled as "generic ASGI request" (#1277) by @BeryJu

Various fixes & improvements

• Add support for cron jobs in ARQ integration (#2088) by @lewazo
• Backpressure handling prototype (#2189) by @sl0thentr0py
• Add "replay" context to event payload (#2234) by @antonpirker
• Update test Django app to be compatible for Django 4.x (#1794) by @DilLip-Chowdary-Codes

Various fixes & improvements

• Add Starlette/FastAPI template tag for adding Sentry tracing information (#2225) by @antonpirker
  • By adding {{ sentry_trace_meta }} to your Starlette/FastAPI Jinja2 templates we will include Sentry trace information as a meta tag in the rendered HTML to allow your frontend to pick up and continue the trace started in the backend.
• Fixed generation of baggage when a DSC is already in propagation context (#2232) by @antonpirker
• Handle explicitly passing None for trace_configs in aiohttp (#2230) by @Harmon758
• Support newest Starlette versions (#2227) by @antonpirker

Various fixes & improvements

• Support for SQLAlchemy 2.0 (#2200) by @antonpirker

• Add instrumentation of aiohttp client requests (#1761) by @md384

• Add Django template tag for adding Sentry tracing information (#2222) by @antonpirker

  • By adding {{ sentry_trace_meta }} to your Django templates we will include Sentry trace information as a meta tag in the rendered HTML to allow your frontend to pick up and continue the trace started in the backend.

• Update Flask HTML meta helper (#2203) by @antonpirker

• Take trace ID always from propagation context (#2209) by @antonpirker

• Fix trace context in event payload (#2205) by @antonpirker

• Use new top level API in trace_propagation_meta (#2202) by @antonpirker

• Do not overwrite existing baggage on outgoing requests (#2191, #2214) by @sentrivana

• Set the transaction/span status from an OTel span (#2115) by @daniil-konovalenko

• Fix propagation of OTel NonRecordingSpan (#2187) by @hartungstenio

• Fix TaskLockedException handling in Huey integration (#2206) by @Zhenay

• Add message format configuration arguments to Loguru integration (#2208) by @Gwill

• Profiling: Add client reports for profiles (#2207) by @Zylphrex

• CI: Fix CI (#2220) by @antonpirker

• Dependencies: Bump checkouts/data-schemas from 7fdde87 to 1b85152 (#2218) by @dependabot

• Dependencies: Bump mypy from 1.3.0 to 1.4.1 (#2194) by @dependabot

• Docs: Change API doc theme (#2210) by @sentrivana

• Docs: Allow (some) autocompletion for top-level API (#2213) by @sentrivana

• Docs: Revert autocomplete hack (#2224) by @sentrivana

Various fixes & improvements

• Tracing without performance (#2136) by @antonpirker
• Load tracing information from environment (#2176) by @antonpirker
• Auto-enable HTTPX integration if HTTPX installed (#2177) by @sentrivana
• Support for SOCKS proxies (#1050) by @Roguelazer
• Wrap parse_url calls in capture_internal_exceptions (#2162) by @sentrivana
• Run 2.7 tests in CI again (#2181) by @sentrivana
• Crons: Do not support sub-minute cron intervals (#2172) by @antonpirker
• Profile: Add function name to profiler frame cache (#2164) by @Zylphrex
• Dependencies: bump checkouts/data-schemas from 0ed3357 to 7fdde87 (#2165) by @dependabot
• Update changelog (#2163) by @sentrivana

Django update (ongoing)

Collections of improvements to our Django integration.

By: @mgaligniana (#1773)

Various fixes & improvements

• Fix parse_url (#2161) by @sentrivana and @antonpirker

  Our URL sanitization used in multiple integrations broke with the recent Python security update. If you started seeing ValueErrors with "'Filtered' does not appear to be an IPv4 or IPv6 address", this release fixes that. See the original issue for more context.

• Better version parsing in integrations (#2152) by @antonpirker

  We now properly support all integration versions that conform to PEP 440. This replaces our naïve version parsing that wouldn't accept versions such as 2.0.0rc1 or 2.0.5.post1.

• Align HTTP status code as span data field http.response.status_code (#2113) by @antonpirker

• Do not encode cached value to determine size (#2143) by @sentrivana

• Fix using unittest.mock whenever available (#1926) by @mgorny

• Fix 2.7 common tests (#2145) by @sentrivana

• Bump actions/stale from 6 to 8 (#1978) by @dependabot

• Bump black from 22.12.0 to 23.3.0 (#1984) by @dependabot

• Bump mypy from 1.2.0 to 1.3.0 (#2110) by @dependabot

• Bump sphinx from 5.3.0 to 7.0.1 (#2112) by @dependabot

Various fixes & improvements

• Support urllib3>=2.0.0 (#2148) by @asottile-sentry

  We're now supporting urllib3's new major version, 2.0.0. If you encounter issues (e.g. some of your dependencies not supporting the new urllib3 version yet) you might consider pinning the urllib3 version to <2.0.0 manually in your project. Check out the the urllib3 migration guide for details.

• Auto-retry tests on failure (#2134) by @sentrivana

• Correct importlib.metadata check in test_modules (#2149) by @asottile-sentry

• Fix distribution name normalization (PEP-0503) (#2144) by @rominf

• Fix functions_to_trace typing (#2141) by @rcmarron

Various fixes & improvements

• New: Celery Beat exclude tasks option (#2130) by @antonpirker

  You can exclude Celery Beat tasks from being auto-instrumented. To do this, add a list of tasks you want to exclude as option exclude_beat_tasks when creating CeleryIntegration. The list can contain simple strings with the full task name, as specified in the Celery Beat schedule, or regular expressions to match multiple tasks.

  For more information, see the documentation for Crons for more information.

  Usage:

      exclude_beat_tasks = [
          "some-task-a",
          "payment-check-.*",
      ]
      sentry_sdk.init(
          dsn='___PUBLIC_DSN___',
          integrations=[
              CeleryIntegration(
                  monitor_beat_tasks=True,
                  exclude_beat_tasks=exclude_beat_tasks,
              ),
          ],
      )
  

  In this example the task some-task-a and all tasks with a name starting with payment-check- will be ignored.

• New: Add support for ExceptionGroups (#2025) by @antonpirker

  Note: If running Self-Hosted Sentry, you should wait to adopt this SDK update until after updating to the 23.6.0 (est. June 2023) release of Sentry. Updating early will not break anything, but you will not get the full benefit of the Exception Groups improvements to issue grouping that were added to the Sentry backend.

• Prefer importlib.metadata over pkg_resources if available (#2081) by @sentrivana

• Work with a copy of request, vars in the event (#2125) by @sentrivana

• Pinned version of dependency that broke the build (#2133) by @antonpirker

Various fixes & improvements

• Disable Django Cache spans by default. (#2120) by @antonpirker

Various fixes & improvements

• New: Add loguru integration (#1994) by @PerchunPak

  Check the documentation for more information.

  Usage:

  from loguru import logger
  import sentry_sdk
  from sentry_sdk.integrations.loguru import LoguruIntegration
  
  sentry_sdk.init(
      dsn="___PUBLIC_DSN___",
      integrations=[
          LoguruIntegration(),
      ],
  )
  
  logger.debug("I am ignored")
  logger.info("I am a breadcrumb")
  logger.error("I am an event", extra=dict(bar=43))
  logger.exception("An exception happened")
  

  • An error event with the message "I am an event" will be created.
  • "I am a breadcrumb" will be attached as a breadcrumb to that event.
  • bar will end up in the extra attributes of that event.
  • "An exception happened" will send the current exception from sys.exc_info() with the stack trace to Sentry. If there's no exception, the current stack will be attached.
  • The debug message "I am ignored" will not be captured by Sentry. To capture it, set level to DEBUG or lower in LoguruIntegration.

• Do not truncate request body if request_bodies is "always" (#2092) by @sentrivana

• Fixed Celery headers for Beat auto-instrumentation (#2102) by @antonpirker

• Add db.operation to Redis and MongoDB spans (#2089) by @antonpirker

• Make sure we're importing redis the library (#2106) by @sentrivana

• Add include_source_context option (#2020) by @farhat-nawaz and @sentrivana

• Import Markup from markupsafe (#2047) by @rco-ableton

• Fix __qualname__ missing attribute in asyncio integration (#2105) by @sl0thentr0py

• Remove relay extension from AWS Layer (#2068) by @sl0thentr0py

• Add a note about pip freeze to the bug template (#2103) by @sentrivana

Various fixes & improvements

• Fix: Django caching spans when using keyword arguments (#2086) by @antonpirker
• Fix: Duration in Celery Beat tasks monitoring (#2087) by @antonpirker
• Fix: Docstrings of SPANDATA (#2084) by @antonpirker

Various fixes & improvements

• Fix: Handle a list of keys (not just a single key) in Django cache spans (#2082) by @antonpirker

Various fixes & improvements

• Add cache.hit and cache.item_size to Django (#2057) by @antonpirker

  Note: This will add spans for all requests to the caches configured in Django. This will probably add some overhead to your server an also add multiple spans to your performance waterfall diagrams. If you do not want this, you can disable this feature in the DjangoIntegration:

  sentry_sdk.init(
      dsn="...",
      integrations=[
          DjangoIntegration(cache_spans=False),
      ]
  )
  

• Use http.method instead of method (#2054) by @AbhiPrasad

• Handle non-int exc.status_code in Starlette (#2075) by @sentrivana

• Handle SQLAlchemy engine.name being bytes (#2074) by @sentrivana

• Fix KeyError in capture_checkin if SDK is not initialized (#2073) by @antonpirker

• Use functools.wrap for ThreadingIntegration patches to fix attributes (#2080) by @EpicWink

• Pin urllib3 to <2.0.0 for now (#2069) by @sl0thentr0py

Various fixes & improvements

• Do not send monitor_config when unset (#2058) by @evanpurkhiser
• Add db.system span data (#2040, #2042) by @antonpirker
• Fix memory leak in profiling (#2049) by @Zylphrex
• Fix crash loop when returning none in before_send (#2045) by @sentrivana

Various fixes & improvements

• Better handling of redis span/breadcrumb data (#2033) by @antonpirker

  Note: With this release we will limit the description of redis db spans and the data in breadcrumbs represting redis db operations to 1024 characters.

  This can can lead to truncated data. If you do not want this there is a new parameter max_data_size in RedisIntegration. You can set this to None for disabling trimming.

  Example for disabling trimming of redis commands in spans or breadcrumbs:

  sentry_sdk.init(
    integrations=[
      RedisIntegration(max_data_size=None),
    ]
  )
  

  Example for custom trim size of redis commands in spans or breadcrumbs:

  sentry_sdk.init(
    integrations=[
      RedisIntegration(max_data_size=50),
    ]
  )`
  
  

• Add db.system to redis and SQLAlchemy db spans (#2037, #2038, #2039) (#2037) by @AbhiPrasad

• Upgraded linting tooling (#2026) by @antonpirker

• Made code more resilient. (#2031) by @antonpirker

Various fixes & improvements

• Send all events to /envelope endpoint when tracing is enabled (#2009) by @antonpirker

  Note: If you’re self-hosting Sentry 9, you need to stay in the previous version of the SDK or update your self-hosted to at least 20.6.0

• Profiling: Remove profile context from SDK (#2013) by @Zylphrex

• Profiling: Additionl performance improvements to the profiler (#1991) by @Zylphrex

• Fix: Celery Beat monitoring without restarting the Beat process (#2001) by @antonpirker

• Fix: Using the Codecov uploader instead of deprecated python package (#2011) by @antonpirker

• Fix: Support for Quart (#2003)` (#2003) by @antonpirker

Various fixes & improvements

• Make auto monitoring beat update support Celery 4 and 5 (#1989) by @antonpirker

Various fixes & improvements

• New: Celery Beat auto monitoring (#1967) by @antonpirker

  The CeleryIntegration can now also monitor your Celery Beat scheduled tasks automatically using the new Crons feature of Sentry.

  To learn more see our Celery Beat Auto Discovery documentation.

  Usage:

  from celery import Celery, signals
  from celery.schedules import crontab
  
  import sentry_sdk
  from sentry_sdk.integrations.celery import CeleryIntegration
  
  
  app = Celery('tasks', broker='...')
  app.conf.beat_schedule = {
      'set-in-beat-schedule': {
          'task': 'tasks.some_important_task',
          'schedule': crontab(...),
      },
  }
  
  
  @signals.celeryd_init.connect
  def init_sentry(**kwargs):
      sentry_sdk.init(
          dsn='...',
          integrations=[CeleryIntegration(monitor_beat_tasks=True)],  # 👈 here
          environment="local.dev.grace",
          release="v1.0",
      )
  

  This will auto detect all schedules tasks in your beat_schedule and will monitor them with Sentry Crons.

• New: gRPC integration (#1911) by @hossein-raeisi

  The gRPC integration instruments all incoming requests and outgoing unary-unary, unary-stream grpc requests using grpcio channels.

  To learn more see our gRPC Integration documentation.

  On the server:

  import grpc
  from sentry_sdk.integrations.grpc.server import ServerInterceptor
  
  
  server = grpc.server(
      thread_pool=...,
      interceptors=[ServerInterceptor()],
  )
  

  On the client:

  import grpc
  from sentry_sdk.integrations.grpc.client import ClientInterceptor
  
  
  with grpc.insecure_channel("example.com:12345") as channel:
      channel = grpc.intercept_channel(channel, *[ClientInterceptor()])
  
  

• New: socket integration (#1911) by @hossein-raeisi

  Use this integration to create spans for DNS resolves (socket.getaddrinfo()) and connection creations (socket.create_connection()).

  To learn more see our Socket Integration documentation.

  Usage:

  import sentry_sdk
  from sentry_sdk.integrations.socket import SocketIntegration
  sentry_sdk.init(
      dsn="___PUBLIC_DSN___",
      integrations=[
          SocketIntegration(),
      ],
  )
  

• Fix: Do not trim span descriptions. (#1983) by @antonpirker

Various fixes & improvements

• New: Implement EventScrubber (#1943) by @sl0thentr0py

  To learn more see our Scrubbing Sensitive Data documentation.

  Add a new EventScrubber class that scrubs certain potentially sensitive interfaces with a DEFAULT_DENYLIST. The default scrubber is automatically run if send_default_pii = False:

  import sentry_sdk
  from sentry_sdk.scrubber import EventScrubber
  sentry_sdk.init(
      # ...
      send_default_pii=False,
      event_scrubber=EventScrubber(),  # this is set by default
  )
  

  You can also pass in a custom denylist to the EventScrubber class and filter additional fields that you want.

  from sentry_sdk.scrubber import EventScrubber, DEFAULT_DENYLIST
  # custom denylist
  denylist = DEFAULT_DENYLIST + ["my_sensitive_var"]
  sentry_sdk.init(
      # ...
      send_default_pii=False,
      event_scrubber=EventScrubber(denylist=denylist),
  )
  

• New: Added new functions_to_trace option for central way of performance instrumentation (#1960) by @antonpirker

  To learn more see our Tracing Options documentation.

  An optional list of functions that should be set up for performance monitoring. For each function in the list, a span will be created when the function is executed.

  functions_to_trace = [
      {"qualified_name": "tests.test_basics._hello_world_counter"},
      {"qualified_name": "time.sleep"},
      {"qualified_name": "collections.Counter.most_common"},
  ]
  
  sentry_sdk.init(
      # ...
      traces_sample_rate=1.0,
      functions_to_trace=functions_to_trace,
  )
  

• Updated denylist to include other widely used cookies/headers (#1972) by @antonpirker

• Forward all sentry- baggage items (#1970) by @cleptric

• Update OSS licensing (#1973) by @antonpirker

• Profiling: Handle non frame types in profiler (#1965) by @Zylphrex

• Tests: Bad arq dependency in tests (#1966) by @Zylphrex

• Better naming (#1962) by @antonpirker

Various fixes & improvements

• New: Monitor Celery Beat tasks with Sentry Cron Monitoring.

  With this feature you can make sure that your Celery beat tasks run at the right time and see if they where successful or not.

  Warning Cron Monitoring is currently in beta. Beta features are still in-progress and may have bugs. We recognize the irony. If you have any questions or feedback, please email us at crons-feedback@sentry.io, reach out via Discord (#cronjobs), or open an issue.

  Usage:

  # File: tasks.py
  
  from celery import Celery, signals
  from celery.schedules import crontab
  
  import sentry_sdk
  from sentry_sdk.crons import monitor
  from sentry_sdk.integrations.celery import CeleryIntegration
  
  
  # 1. Setup your Celery beat configuration
  
  app = Celery('mytasks', broker='redis://localhost:6379/0')
  app.conf.beat_schedule = {
      'set-in-beat-schedule': {
          'task': 'tasks.tell_the_world',
          'schedule': crontab(hour='10', minute='15'),
          'args': ("in beat_schedule set", ),
      },
  }
  
  
  # 2. Initialize Sentry either in `celeryd_init` or `beat_init` signal.
  
  #@signals.celeryd_init.connect
  @signals.beat_init.connect
  def init_sentry(**kwargs):
      sentry_sdk.init(
          dsn='...',
          integrations=[CeleryIntegration()],
          environment="local.dev.grace",
          release="v1.0.7-a1",
      )
  
  
  # 3. Link your Celery task to a Sentry Cron Monitor
  
  @app.task
  @monitor(monitor_slug='3b861d62-ff82-4aa0-9cd6-b2b6403bd0cf')
  def tell_the_world(msg):
      print(msg)
  

• New: Add decorator for Sentry tracing (#1089) by @ynouri

  This allows you to use a decorator to setup custom performance instrumentation.

  To learn more see Custom Instrumentation.

  Usage: Just add the new decorator to your function, and a span will be created for it:

  import sentry_sdk
  
  @sentry_sdk.trace
  def my_complex_function():
    # do stuff
    ...
  

• Make Django signals tracing optional (#1929) by @antonpirker

  See the Django Guide to learn more.

• Deprecated with_locals in favor of include_local_variables (#1924) by @antonpirker

• Added top level API to get current span (#1954) by @antonpirker

• Profiling: Add profiler options to init (#1947) by @Zylphrex

• Profiling: Set active thread id for quart (#1830) by @Zylphrex

• Fix: Update get_json function call for werkzeug 2.1.0+ (#1939) by @michielderoos

• Fix: Returning the tasks result. (#1931) by @antonpirker

• Fix: Rename MYPY to TYPE_CHECKING (#1934) by @untitaker

• Fix: Fix type annotation for ignore_errors in sentry_sdk.init() (#1928) by @tiangolo

• Tests: Start a real http server instead of mocking libs (#1938) by @antonpirker

Various fixes & improvements

• New: Add arq Integration (#1872) by @Zhenay

  This integration will create performance spans when arq jobs will be enqueued and when they will be run. It will also capture errors in jobs and will link them to the performance spans.

  Usage:

  import asyncio
  
  from httpx import AsyncClient
  from arq import create_pool
  from arq.connections import RedisSettings
  
  import sentry_sdk
  from sentry_sdk.integrations.arq import ArqIntegration
  from sentry_sdk.tracing import TRANSACTION_SOURCE_COMPONENT
  
  sentry_sdk.init(
      dsn="...",
      integrations=[ArqIntegration()],
  )
  
  async def download_content(ctx, url):
      session: AsyncClient = ctx['session']
      response = await session.get(url)
      print(f'{url}: {response.text:.80}...')
      return len(response.text)
  
  async def startup(ctx):
      ctx['session'] = AsyncClient()
  
  async def shutdown(ctx):
      await ctx['session'].aclose()
  
  async def main():
      with sentry_sdk.start_transaction(name="testing_arq_tasks", source=TRANSACTION_SOURCE_COMPONENT):
          redis = await create_pool(RedisSettings())
          for url in ('https://facebook.com', 'https://microsoft.com', 'https://github.com', "asdf"
                      ):
              await redis.enqueue_job('download_content', url)
  
  class WorkerSettings:
      functions = [download_content]
      on_startup = startup
      on_shutdown = shutdown
  
  if __name__ == '__main__':
      asyncio.run(main())
  

• Update of Falcon Integration (#1733) by @bartolootrit

• Adding Cloud Resource Context integration (#1882) by @antonpirker

• Profiling: Use the transaction timestamps to anchor the profile (#1898) by @Zylphrex

• Profiling: Add debug logs to profiling (#1883) by @Zylphrex

• Profiling: Start profiler thread lazily (#1903) by @Zylphrex

• Fixed checks for structured http data (#1905) by @antonpirker

• Make set_measurement public api and remove experimental status (#1909) by @sl0thentr0py

• Add trace_propagation_targets option (#1916) by @antonpirker

• Add enable_tracing to default traces_sample_rate to 1.0 (#1900) by @sl0thentr0py

• Remove deprecated tracestate (#1907) by @sl0thentr0py

• Sanitize URLs in Span description and breadcrumbs (#1876) by @antonpirker

• Mechanism should default to true unless set explicitly (#1889) by @sl0thentr0py

• Better setting of in-app in stack frames (#1894) by @antonpirker

• Add workflow to test gevent (#1870) by @Zylphrex

• Updated outdated HTTPX test matrix (#1917) by @antonpirker

• Switch to MIT license (#1908) by @cleptric

Various fixes & improvements

• New: Add Huey Integration (#1555) by @Zhenay

  This integration will create performance spans when Huey tasks will be enqueued and when they will be executed.

  Usage:

  Task definition in demo.py:

  import time
  
  from huey import SqliteHuey, crontab
  
  import sentry_sdk
  from sentry_sdk.integrations.huey import HueyIntegration
  
  sentry_sdk.init(
      dsn="...",
      integrations=[
          HueyIntegration(),
      ],
      traces_sample_rate=1.0,
  )
  
  huey = SqliteHuey(filename='/tmp/demo.db')
  
  @huey.task()
  def add_numbers(a, b):
      return a + b
  

  Running the tasks in run.py:

  from demo import add_numbers, flaky_task, nightly_backup
  
  import sentry_sdk
  from sentry_sdk.integrations.huey import HueyIntegration
  from sentry_sdk.tracing import TRANSACTION_SOURCE_COMPONENT, Transaction
  
  
  def main():
      sentry_sdk.init(
          dsn="...",
          integrations=[
              HueyIntegration(),
          ],
          traces_sample_rate=1.0,
      )
  
      with sentry_sdk.start_transaction(name="testing_huey_tasks", source=TRANSACTION_SOURCE_COMPONENT):
          r = add_numbers(1, 2)
  
  if __name__ == "__main__":
      main()
  

• Profiling: Do not send single sample profiles (#1879) by @Zylphrex

• Profiling: Add additional test coverage for profiler (#1877) by @Zylphrex

• Profiling: Always use builtin time.sleep (#1869) by @Zylphrex

• Profiling: Defaul in_app decision to None (#1855) by @Zylphrex

• Profiling: Remove use of threading.Event (#1864) by @Zylphrex

• Profiling: Enable profiling on all transactions (#1797) by @Zylphrex

• FastAPI: Fix check for Starlette in FastAPI integration (#1868) by @antonpirker

• Flask: Do not overwrite default for username with email address in FlaskIntegration (#1873) by @homeworkprod

• Tests: Add py3.11 to test-common (#1871) by @Zylphrex

• Fix: Don't log whole event in before_send / event_processor drops (#1863) by @sl0thentr0py

Various fixes & improvements

• Add before_send_transaction (#1840) by @antonpirker

  Adds a hook (similar to before_send) that is called for all transaction events (performance releated data).

  Usage:

    import sentry_sdk
  
    def strip_sensitive_data(event, hint):
        # modify event here (or return `None` if you want to drop the event entirely)
        return event
  
    sentry_sdk.init(
        # ...
        before_send_transaction=strip_sensitive_data,
    )
  

  See also: https://docs.sentry.io/platforms/python/configuration/filtering/#using-platformidentifier-namebefore-send-transaction-

• Django: Always remove values of Django session related cookies. (#1842) by @antonpirker

• Profiling: Enable profiling for ASGI frameworks (#1824) by @Zylphrex

• Profiling: Better gevent support (#1822) by @Zylphrex

• Profiling: Add profile context to transaction (#1860) by @Zylphrex

• Profiling: Use co_qualname in python 3.11 (#1831) by @Zylphrex

• OpenTelemetry: fix Use dict for sentry-trace context instead of tuple (#1847) by @AbhiPrasad

• OpenTelemetry: fix extra dependency (#1825) by @bernardotorres

• OpenTelemetry: fix NoOpSpan updates scope (#1834) by @Zylphrex

• OpenTelemetry: Make sure to noop when there is no DSN (#1852) by @antonpirker

• FastAPI: Fix middleware being patched multiple times (#1841) by @JohnnyDeuss

• Starlette: Avoid import of pkg_resource with Starlette integration (#1836) by @mgu

• Removed code coverage target (#1862) by @antonpirker

Various fixes & improvements

• Add Starlite integration (#1748) by @gazorby

  Adding support for the Starlite framework. Unhandled errors are captured. Performance spans for Starlite middleware are also captured. Thanks @gazorby for the great work!

  Usage:

  from starlite import Starlite, get
  
  import sentry_sdk
  from sentry_sdk.integrations.starlite import StarliteIntegration
  
  sentry_sdk.init(
      dsn="...",
      traces_sample_rate=1.0,
      integrations=[
          StarliteIntegration(),
      ],
  )
  
  @get("/")
  def hello_world() -> dict[str, str]:
      """Keeping the tradition alive with hello world."""
      bla = 1/0  # causing an error
      return {"hello": "world"}
  
  app = Starlite(route_handlers=[hello_world])
  

• Profiling: Remove sample buffer from profiler (#1791) by @Zylphrex

• Profiling: Performance tweaks to profile sampler (#1789) by @Zylphrex

• Add span for Django SimpleTemplateResponse rendering (#1818) by @chdsbd

• Use @wraps for Django Signal receivers (#1815) by @meanmail

• Add enqueued_at and started_at to rq job extra (#1024) by @kruvasyan

• Remove sanic v22 pin (#1819) by @sl0thentr0py

• Add support for byterray and memoryview built-in types (#1833) by @Tarty

• Handle "rc" in SQLAlchemy version. (#1812) by @peterschutt

• Doc: Use .venv (not .env) as a virtual env location in CONTRIBUTING.md (#1790) by @tonyo

• Auto publish to internal pypi on release (#1823) by @asottile-sentry

• Added Python 3.11 to test suite (#1795) by @antonpirker

• Update test/linting dependencies (#1801) by @antonpirker

• Deps: bump sphinx from 5.2.3 to 5.3.0 (#1686) by @dependabot

Various fixes & improvements

• Link errors to OTel spans (#1787) by @antonpirker