Merge branch 'master' into sentry-sdk-2.0

Author: antonpirker

CHANGELOG.md

@@ -83,6 +83,58 @@
 - 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`.
 
+## 1.41.0
+
+### 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:
+
+  ```python
+  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:
+
+  ```python
+  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
+
 ## 1.40.6
 
 ### Various fixes & improvements

checkouts/data-schemas

@@ -2,7 +2,7 @@ mypy
 black
 flake8==5.0.4  # flake8 depends on pyflakes>=3.0.0 and this dropped support for Python 2 "# type:" comments
 types-certifi
-types-protobuf==4.24.0.20240129  # newer raises an error on mypy sentry_sdk
+types-protobuf==4.24.0.20240302  # newer raises an error on mypy sentry_sdk
 types-redis
 types-setuptools
 pymongo # There is no separate types module.

linter-requirements.txt

@@ -124,6 +124,12 @@ def _get_options(*args, **kwargs):
     if rv["event_scrubber"] is None:
         rv["event_scrubber"] = EventScrubber()
 
+    if rv["socket_options"] and not isinstance(rv["socket_options"], list):
+        logger.warning(
+            "Ignoring socket_options because of unexpected format. See urllib3.HTTPConnection.socket_options for the expected format."
+        )
+        rv["socket_options"] = None
+
     return rv
 
 

sentry_sdk/client.py

@@ -27,6 +27,7 @@ class EndpointType(Enum):
     from typing import Dict
     from typing import Any
     from typing import Sequence
+    from typing import Tuple
     from typing_extensions import TypedDict
 
     from sentry_sdk.integrations import Integration
@@ -270,6 +271,7 @@ def __init__(
         https_proxy=None,  # type: Optional[str]
         ignore_errors=[],  # type: Sequence[Union[type, str]]  # noqa: B006
         max_request_body_size="medium",  # type: str
+        socket_options=None,  # type: Optional[List[Tuple[int, int, int | bytes]]]
         before_send=None,  # type: Optional[EventProcessor]
         before_breadcrumb=None,  # type: Optional[BreadcrumbProcessor]
         debug=None,  # type: Optional[bool]

sentry_sdk/consts.py

@@ -7,12 +7,15 @@
 import time
 from datetime import datetime, timedelta, timezone
 from collections import defaultdict
+import certifi
+import urllib3
 from urllib.request import getproxies
 
 from sentry_sdk.consts import EndpointType
 from sentry_sdk.utils import Dsn, logger, capture_internal_exceptions
 from sentry_sdk.worker import BackgroundWorker
 from sentry_sdk.envelope import Envelope, Item, PayloadRef
+
 from sentry_sdk._types import TYPE_CHECKING
 
 if TYPE_CHECKING:
@@ -417,12 +420,17 @@ def _send_envelope(
 
     def _get_pool_options(self, ca_certs):
         # type: (Optional[Any]) -> Dict[str, Any]
-        return {
+        options = {
             "num_pools": self._num_pools,
             "cert_reqs": "CERT_REQUIRED",
             "ca_certs": ca_certs or certifi.where(),
         }
 
+        if self.options["socket_options"]:
+            options["socket_options"] = self.options["socket_options"]
+
+        return options
+
     def _in_no_proxy(self, parsed_dsn):
         # type: (Dsn) -> bool
         no_proxy = getproxies().get("no")

sentry_sdk/transport.py

@@ -240,7 +240,7 @@ def run_lambda_function(
             FunctionName=full_fn_name,
         )
         print(
-            f"Lambda function {full_fn_name} in AWS already existing, taking it (and do not create a local one)"
+            "Lambda function in AWS already existing, taking it (and do not create a local one)"
         )
     except client.exceptions.ResourceNotFoundException:
         function_exists_in_aws = False
@@ -251,14 +251,9 @@ def run_lambda_function(
         dir_already_existing = os.path.isdir(base_dir)
 
         if dir_already_existing:
-            print(
-                f"Local Lambda function directory ({base_dir}) already exists, skipping creation"
-            )
+            print("Local Lambda function directory already exists, skipping creation")
 
         if not dir_already_existing:
-            print(
-                f"Creating Lambda function package ({full_fn_name}) locally in directory {base_dir}"
-            )
             os.mkdir(base_dir)
             _create_lambda_package(
                 base_dir, code, initial_handler, layer, syntax_check, subprocess_kwargs
@@ -321,10 +316,9 @@ def clean_up():
 
                 waiter = client.get_waiter("function_active_v2")
                 waiter.wait(FunctionName=full_fn_name)
-                print(f"Created Lambda function in AWS: {full_fn_name}")
         except client.exceptions.ResourceConflictException:
             print(
-                f"Lambda function ({full_fn_name}) already existing in AWS, this is fine, we will just invoke it."
+                "Lambda function already exists, this is fine, we will just invoke it."
             )
 
     response = client.invoke(

tests/integrations/aws_lambda/client.py

@@ -650,6 +650,9 @@ def test_handler(event, context):
     assert response["Payload"]["AssertionError raised"] is False
 
 
+@pytest.mark.xfail(
+    reason="The limited log output we depend on is being clogged by a new warning"
+)
 def test_serverless_no_code_instrumentation(run_lambda_function):
     """
     Test that ensures that just by adding a lambda layer containing the
@@ -694,6 +697,9 @@ def test_handler(event, context):
         assert "sentry_handler" in response["LogResult"][3].decode("utf-8")
 
 
+@pytest.mark.xfail(
+    reason="The limited log output we depend on is being clogged by a new warning"
+)
 def test_error_has_new_trace_context_performance_enabled(run_lambda_function):
     envelope_items, _ = run_lambda_function(
         LAMBDA_PRELUDE
@@ -756,6 +762,9 @@ def test_handler(event, context):
     )
 
 
+@pytest.mark.xfail(
+    reason="The limited log output we depend on is being clogged by a new warning"
+)
 def test_error_has_existing_trace_context_performance_enabled(run_lambda_function):
     trace_id = "471a43a4192642f0b136d5159a501701"
     parent_span_id = "6e8f22c393e68f19"

tests/integrations/aws_lambda/test_aws.py

@@ -2,6 +2,7 @@
 import pickle
 import gzip
 import io
+import socket
 from collections import namedtuple
 from datetime import datetime, timedelta, timezone
 from unittest import mock
@@ -150,6 +151,19 @@ def test_transport_num_pools(make_client, num_pools, expected_num_pools):
     assert options["num_pools"] == expected_num_pools
 
 
+def test_socket_options(make_client):
+    socket_options = [
+        (socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1),
+        (socket.SOL_TCP, socket.TCP_KEEPINTVL, 10),
+        (socket.SOL_TCP, socket.TCP_KEEPCNT, 6),
+    ]
+
+    client = make_client(socket_options=socket_options)
+
+    options = client.transport._get_pool_options([])
+    assert options["socket_options"] == socket_options
+
+
 def test_transport_infinite_loop(capturing_server, request, make_client):
     client = make_client(
         debug=True,
@@ -221,7 +235,7 @@ def test_parse_rate_limits(input, expected):
     assert dict(_parse_rate_limits(input, now=NOW)) == expected
 
 
-def test_simple_rate_limits(capturing_server, capsys, caplog, make_client):
+def test_simple_rate_limits(capturing_server, make_client):
     client = make_client()
     capturing_server.respond_with(code=429, headers={"Retry-After": "4"})
 
@@ -243,7 +257,7 @@ def test_simple_rate_limits(capturing_server, capsys, caplog, make_client):
 
 @pytest.mark.parametrize("response_code", [200, 429])
 def test_data_category_limits(
-    capturing_server, capsys, caplog, response_code, make_client, monkeypatch
+    capturing_server, response_code, make_client, monkeypatch
 ):
     client = make_client(send_client_reports=False)
 
@@ -290,7 +304,7 @@ def record_lost_event(reason, data_category=None, item=None):
 
 @pytest.mark.parametrize("response_code", [200, 429])
 def test_data_category_limits_reporting(
-    capturing_server, capsys, caplog, response_code, make_client, monkeypatch
+    capturing_server, response_code, make_client, monkeypatch
 ):
     client = make_client(send_client_reports=True)
 
@@ -374,7 +388,7 @@ def intercepting_fetch(*args, **kwargs):
 
 @pytest.mark.parametrize("response_code", [200, 429])
 def test_complex_limits_without_data_category(
-    capturing_server, capsys, caplog, response_code, make_client
+    capturing_server, response_code, make_client
 ):
     client = make_client()
     capturing_server.respond_with(