chore(similarity): Add logging for over 30 system frames

src/sentry/grouping/ingest/seer.py

@@ -187,7 +187,11 @@ def get_seer_similar_issues(
     should go in (if any), or None if no neighbor was near enough.
     """
     event_hash = event.get_primary_hash()
-    stacktrace_string = get_stacktrace_string(get_grouping_info_from_variants(variants))
+    # Temporarily add project id to this for logging purposes
+    # TODO: Remove when grouping.similarity.over_threshold_system_only_frames is removed
+    grouping_info = get_grouping_info_from_variants(variants)
+    grouping_info["project_id"] = event.project.id
+    stacktrace_string = get_stacktrace_string(grouping_info)
     exception_type = get_path(event.data, "exception", "values", -1, "type")
 
     request_data: SimilarIssuesEmbeddingsRequest = {

src/sentry/seer/similarity/utils.py

@@ -179,18 +179,22 @@ def get_stacktrace_string(data: dict[str, Any]) -> str:
     html_frame_count = 0  # for a temporary metric
     stacktrace_str = ""
     found_non_snipped_context_line = False
+    is_frames_truncated = False
 
     metrics.distribution("seer.grouping.exceptions.length", len(exceptions))
 
     def _process_frames(frames: list[dict[str, Any]]) -> list[str]:
         nonlocal frame_count
         nonlocal html_frame_count
         nonlocal found_non_snipped_context_line
+        nonlocal is_frames_truncated
         frame_strings = []
 
         contributing_frames = [
             frame for frame in frames if frame.get("id") == "frame" and frame.get("contributes")
         ]
+        if len(contributing_frames) + frame_count > MAX_FRAME_COUNT:
+            is_frames_truncated = True
         contributing_frames = _discard_excess_frames(
             contributing_frames, MAX_FRAME_COUNT, frame_count
         )
@@ -287,6 +291,14 @@ def _process_frames(frames: list[dict[str, Any]]) -> list[str]:
         },
     )
 
+    if is_frames_truncated and not app_hash:
+        logger_extra = {
+            "project_id": data.get("project_id", ""),
+            "hash": system_hash,
+            "stacktrace_str": stacktrace_str.strip(),
+        }
+        logger.info("grouping.similarity.over_threshold_system_only_frames", extra=logger_extra)
+
     return stacktrace_str.strip()
 
 

src/sentry/tasks/embeddings_grouping/utils.py

@@ -355,6 +355,9 @@ def get_events_from_nodestore(
         event._project_cache = project
         if event and event.data and event_content_has_stacktrace(event):
             grouping_info = get_grouping_info(None, project=project, event=event)
+            # Temporarily add project id to this for logging purposes
+            # TODO: Remove when grouping.similarity.over_threshold_system_only_frames is removed
+            grouping_info["project_id"] = project.id
             stacktrace_string = get_stacktrace_string(grouping_info)
             if stacktrace_string == "":
                 invalid_event_group_ids.append(group_id)

tests/sentry/grouping/ingest/test_seer.py

@@ -1,12 +1,14 @@
 from dataclasses import asdict
 from time import time
-from unittest.mock import MagicMock, patch
+from unittest.mock import MagicMock, Mock, patch
 
 from sentry.conf.server import SEER_SIMILARITY_MODEL_VERSION
 from sentry.eventstore.models import Event
+from sentry.grouping.grouping_info import get_grouping_info_from_variants
 from sentry.grouping.ingest.seer import get_seer_similar_issues, should_call_seer_for_grouping
 from sentry.models.grouphash import GroupHash
 from sentry.seer.similarity.types import SeerSimilarIssueData
+from sentry.seer.similarity.utils import MAX_FRAME_COUNT
 from sentry.testutils.cases import TestCase
 from sentry.testutils.helpers.eventprocessing import save_new_event
 from sentry.testutils.helpers.options import override_options
@@ -261,3 +263,47 @@ def test_returns_no_grouphash_and_empty_metadata_if_no_similar_group_found(self)
                 expected_metadata,
                 None,
             )
+
+    @patch("sentry.seer.similarity.utils.logger")
+    def test_too_many_only_system_frames(self, mock_logger: Mock) -> None:
+        type = "FailedToFetchError"
+        value = "Charlie didn't bring the ball back"
+        context_line = f"raise {type}('{value}')"
+        new_event = Event(
+            project_id=self.project.id,
+            event_id="22312012112120120908201304152013",
+            data={
+                "title": f"{type}('{value}')",
+                "exception": {
+                    "values": [
+                        {
+                            "type": type,
+                            "value": value,
+                            "stacktrace": {
+                                "frames": [
+                                    {
+                                        "function": f"play_fetch_{i}",
+                                        "filename": f"dogpark{i}.py",
+                                        "context_line": context_line,
+                                    }
+                                    for i in range(MAX_FRAME_COUNT + 1)
+                                ]
+                            },
+                        }
+                    ]
+                },
+                "platform": "python",
+            },
+        )
+        variants = new_event.get_grouping_variants()
+        get_seer_similar_issues(new_event, variants)
+
+        grouping_info = get_grouping_info_from_variants(variants)
+        mock_logger.info.assert_called_with(
+            "grouping.similarity.over_threshold_system_only_frames",
+            extra={
+                "project_id": self.project.id,
+                "hash": grouping_info["system"]["hash"],
+                "stacktrace_str": "FailedToFetchError: Charlie didn't bring the ball back\n  File \"dogpark1.py\", function play_fetch_1\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark2.py\", function play_fetch_2\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark3.py\", function play_fetch_3\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark4.py\", function play_fetch_4\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark5.py\", function play_fetch_5\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark6.py\", function play_fetch_6\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark7.py\", function play_fetch_7\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark8.py\", function play_fetch_8\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark9.py\", function play_fetch_9\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark10.py\", function play_fetch_10\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark11.py\", function play_fetch_11\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark12.py\", function play_fetch_12\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark13.py\", function play_fetch_13\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark14.py\", function play_fetch_14\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark15.py\", function play_fetch_15\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark16.py\", function play_fetch_16\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark17.py\", function play_fetch_17\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark18.py\", function play_fetch_18\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark19.py\", function play_fetch_19\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark20.py\", function play_fetch_20\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark21.py\", function play_fetch_21\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark22.py\", function play_fetch_22\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark23.py\", function play_fetch_23\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark24.py\", function play_fetch_24\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark25.py\", function play_fetch_25\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark26.py\", function play_fetch_26\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark27.py\", function play_fetch_27\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark28.py\", function play_fetch_28\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark29.py\", function play_fetch_29\n    raise FailedToFetchError('Charlie didn't bring the ball back')\n  File \"dogpark30.py\", function play_fetch_30\n    raise FailedToFetchError('Charlie didn't bring the ball back')",
+            },
+        )

tests/sentry/seer/similarity/test_utils.py

@@ -1,11 +1,13 @@
 import copy
 from collections.abc import Callable
 from typing import Any, Literal, cast
+from unittest.mock import patch
 from uuid import uuid1
 
 from sentry.eventstore.models import Event
 from sentry.seer.similarity.utils import (
     BASE64_ENCODED_PREFIXES,
+    MAX_FRAME_COUNT,
     SEER_ELIGIBLE_PLATFORMS,
     _is_snipped_context_line,
     event_content_is_seer_eligible,
@@ -710,7 +712,51 @@ def test_no_app_no_system(self):
         stacktrace_str = get_stacktrace_string(data)
         assert stacktrace_str == ""
 
-    def test_over_30_contributing_frames(self):
+    @patch("sentry.seer.similarity.utils.logger")
+    def test_too_many_system_frames_single_exception(self, mock_logger):
+        data_system = copy.deepcopy(self.BASE_APP_DATA)
+        data_system["system"] = data_system.pop("app")
+        data_system["system"]["component"]["values"][0]["values"][0][
+            "values"
+        ] += self.create_frames(MAX_FRAME_COUNT + 1, True)
+        data_system["project_id"] = self.project.id
+
+        get_stacktrace_string(data_system)
+
+        mock_logger.info.assert_called_with(
+            "grouping.similarity.over_threshold_system_only_frames",
+            extra={
+                "project_id": self.project.id,
+                "hash": data_system["system"]["hash"],
+                "stacktrace_str": 'ZeroDivisionError: division by zero\n  File "hello.py", function hello_there\n    test = 2!\n  File "hello.py", function hello_there\n    test = 3!\n  File "hello.py", function hello_there\n    test = 4!\n  File "hello.py", function hello_there\n    test = 5!\n  File "hello.py", function hello_there\n    test = 6!\n  File "hello.py", function hello_there\n    test = 7!\n  File "hello.py", function hello_there\n    test = 8!\n  File "hello.py", function hello_there\n    test = 9!\n  File "hello.py", function hello_there\n    test = 10!\n  File "hello.py", function hello_there\n    test = 11!\n  File "hello.py", function hello_there\n    test = 12!\n  File "hello.py", function hello_there\n    test = 13!\n  File "hello.py", function hello_there\n    test = 14!\n  File "hello.py", function hello_there\n    test = 15!\n  File "hello.py", function hello_there\n    test = 16!\n  File "hello.py", function hello_there\n    test = 17!\n  File "hello.py", function hello_there\n    test = 18!\n  File "hello.py", function hello_there\n    test = 19!\n  File "hello.py", function hello_there\n    test = 20!\n  File "hello.py", function hello_there\n    test = 21!\n  File "hello.py", function hello_there\n    test = 22!\n  File "hello.py", function hello_there\n    test = 23!\n  File "hello.py", function hello_there\n    test = 24!\n  File "hello.py", function hello_there\n    test = 25!\n  File "hello.py", function hello_there\n    test = 26!\n  File "hello.py", function hello_there\n    test = 27!\n  File "hello.py", function hello_there\n    test = 28!\n  File "hello.py", function hello_there\n    test = 29!\n  File "hello.py", function hello_there\n    test = 30!\n  File "hello.py", function hello_there\n    test = 31!',
+            },
+        )
+
+    @patch("sentry.seer.similarity.utils.logger")
+    def test_too_many_system_frames_chained_exception(self, mock_logger):
+        data_system = copy.deepcopy(self.CHAINED_APP_DATA)
+        data_system["system"] = data_system.pop("app")
+        data_system["project_id"] = self.project.id
+        # Split MAX_FRAME_COUNT across the two exceptions
+        data_system["system"]["component"]["values"][0]["values"][0]["values"][0][
+            "values"
+        ] += self.create_frames(MAX_FRAME_COUNT // 2, True)
+        data_system["system"]["component"]["values"][0]["values"][1]["values"][0][
+            "values"
+        ] += self.create_frames(MAX_FRAME_COUNT // 2, True)
+
+        get_stacktrace_string(data_system)
+
+        mock_logger.info.assert_called_with(
+            "grouping.similarity.over_threshold_system_only_frames",
+            extra={
+                "project_id": self.project.id,
+                "hash": data_system["system"]["hash"],
+                "stacktrace_str": 'Exception: Catch divide by zero error\n  File "python_onboarding.py", function <module>\n    divide_by_zero()\n  File "python_onboarding.py", function divide_by_zero\n    raise Exception("Catch divide by zero error")\n  File "hello.py", function hello_there\n    test = 1!\n  File "hello.py", function hello_there\n    test = 2!\n  File "hello.py", function hello_there\n    test = 3!\n  File "hello.py", function hello_there\n    test = 4!\n  File "hello.py", function hello_there\n    test = 5!\n  File "hello.py", function hello_there\n    test = 6!\n  File "hello.py", function hello_there\n    test = 7!\n  File "hello.py", function hello_there\n    test = 8!\n  File "hello.py", function hello_there\n    test = 9!\n  File "hello.py", function hello_there\n    test = 10!\n  File "hello.py", function hello_there\n    test = 11!\n  File "hello.py", function hello_there\n    test = 12!\n  File "hello.py", function hello_there\n    test = 13!\n  File "hello.py", function hello_there\n    test = 14!\n  File "hello.py", function hello_there\n    test = 15!\nZeroDivisionError: division by zero\n  File "hello.py", function hello_there\n    test = 3!\n  File "hello.py", function hello_there\n    test = 4!\n  File "hello.py", function hello_there\n    test = 5!\n  File "hello.py", function hello_there\n    test = 6!\n  File "hello.py", function hello_there\n    test = 7!\n  File "hello.py", function hello_there\n    test = 8!\n  File "hello.py", function hello_there\n    test = 9!\n  File "hello.py", function hello_there\n    test = 10!\n  File "hello.py", function hello_there\n    test = 11!\n  File "hello.py", function hello_there\n    test = 12!\n  File "hello.py", function hello_there\n    test = 13!\n  File "hello.py", function hello_there\n    test = 14!\n  File "hello.py", function hello_there\n    test = 15!',
+            },
+        )
+
+    def test_too_many_in_app_contributing_frames(self):
         """Check that when there are over 30 contributing frames, the last 30 are included."""
 
         data_frames = copy.deepcopy(self.BASE_APP_DATA)