|
| 1 | +from dataclasses import asdict |
| 2 | +from typing import Any |
| 3 | +from unittest.mock import MagicMock, patch |
| 4 | + |
| 5 | +from sentry.conf.server import SEER_SIMILARITY_MODEL_VERSION |
| 6 | +from sentry.grouping.ingest.seer import get_seer_similar_issues, should_call_seer_for_grouping |
| 7 | +from sentry.seer.utils import SeerSimilarIssueData |
| 8 | +from sentry.testutils.cases import TestCase |
| 9 | +from sentry.testutils.helpers import Feature |
| 10 | +from sentry.testutils.helpers.eventprocessing import save_new_event |
| 11 | +from sentry.testutils.helpers.features import with_feature |
| 12 | +from sentry.testutils.pytest.mocking import capture_results |
| 13 | +from sentry.utils.types import NonNone |
| 14 | + |
| 15 | + |
| 16 | +class SeerEventManagerGroupingTest(TestCase): |
| 17 | + """Test whether Seer is called during ingest and if so, how the results are used""" |
| 18 | + |
| 19 | + def test_obeys_seer_similarity_flags(self): |
| 20 | + existing_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 21 | + seer_result_data = SeerSimilarIssueData( |
| 22 | + parent_hash=existing_event.get_primary_hash(), |
| 23 | + parent_group_id=NonNone(existing_event.group_id), |
| 24 | + stacktrace_distance=0.01, |
| 25 | + message_distance=0.05, |
| 26 | + should_group=True, |
| 27 | + ) |
| 28 | + metadata_base = { |
| 29 | + "similarity_model_version": SEER_SIMILARITY_MODEL_VERSION, |
| 30 | + "results": [asdict(seer_result_data)], |
| 31 | + } |
| 32 | + |
| 33 | + get_seer_similar_issues_return_values: list[Any] = [] |
| 34 | + |
| 35 | + with ( |
| 36 | + patch( |
| 37 | + "sentry.event_manager.should_call_seer_for_grouping", |
| 38 | + wraps=should_call_seer_for_grouping, |
| 39 | + ) as should_call_seer_spy, |
| 40 | + patch( |
| 41 | + "sentry.event_manager.get_seer_similar_issues", |
| 42 | + wraps=capture_results( |
| 43 | + get_seer_similar_issues, get_seer_similar_issues_return_values |
| 44 | + ), |
| 45 | + ) as get_seer_similar_issues_spy, |
| 46 | + patch( |
| 47 | + "sentry.grouping.ingest.seer.get_similarity_data_from_seer", |
| 48 | + return_value=[seer_result_data], |
| 49 | + ), |
| 50 | + ): |
| 51 | + |
| 52 | + with Feature( |
| 53 | + { |
| 54 | + "projects:similarity-embeddings-metadata": False, |
| 55 | + "projects:similarity-embeddings-grouping": False, |
| 56 | + } |
| 57 | + ): |
| 58 | + new_event = save_new_event({"message": "Adopt don't shop"}, self.project) |
| 59 | + |
| 60 | + # We checked whether to make the call, but didn't go through with it |
| 61 | + assert should_call_seer_spy.call_count == 1 |
| 62 | + assert get_seer_similar_issues_spy.call_count == 0 |
| 63 | + |
| 64 | + # No metadata stored, parent group not used (even though `should_group` is True) |
| 65 | + assert "seer_similarity" not in NonNone(new_event.group).data["metadata"] |
| 66 | + assert "seer_similarity" not in new_event.data |
| 67 | + assert new_event.group_id != existing_event.group_id |
| 68 | + |
| 69 | + should_call_seer_spy.reset_mock() |
| 70 | + get_seer_similar_issues_spy.reset_mock() |
| 71 | + |
| 72 | + with Feature( |
| 73 | + { |
| 74 | + "projects:similarity-embeddings-metadata": True, |
| 75 | + "projects:similarity-embeddings-grouping": False, |
| 76 | + } |
| 77 | + ): |
| 78 | + new_event = save_new_event({"message": "Maisey is silly"}, self.project) |
| 79 | + expected_metadata = {**metadata_base, "request_hash": new_event.get_primary_hash()} |
| 80 | + |
| 81 | + # We checked whether to make the call, and then made it |
| 82 | + assert should_call_seer_spy.call_count == 1 |
| 83 | + assert get_seer_similar_issues_spy.call_count == 1 |
| 84 | + |
| 85 | + # Metadata returned and stored |
| 86 | + assert get_seer_similar_issues_return_values[0][0] == expected_metadata |
| 87 | + assert ( |
| 88 | + NonNone(new_event.group).data["metadata"]["seer_similarity"] |
| 89 | + == expected_metadata |
| 90 | + ) |
| 91 | + assert new_event.data["seer_similarity"] == expected_metadata |
| 92 | + |
| 93 | + # No parent group returned or used (even though `should_group` is True) |
| 94 | + assert get_seer_similar_issues_return_values[0][1] is None |
| 95 | + assert new_event.group_id != existing_event.group_id |
| 96 | + |
| 97 | + should_call_seer_spy.reset_mock() |
| 98 | + get_seer_similar_issues_spy.reset_mock() |
| 99 | + get_seer_similar_issues_return_values.pop() |
| 100 | + |
| 101 | + with Feature( |
| 102 | + { |
| 103 | + "projects:similarity-embeddings-metadata": False, |
| 104 | + "projects:similarity-embeddings-grouping": True, |
| 105 | + } |
| 106 | + ): |
| 107 | + new_event = save_new_event({"message": "Charlie is goofy"}, self.project) |
| 108 | + expected_metadata = {**metadata_base, "request_hash": new_event.get_primary_hash()} |
| 109 | + |
| 110 | + # We checked whether to make the call, and then made it |
| 111 | + assert should_call_seer_spy.call_count == 1 |
| 112 | + assert get_seer_similar_issues_spy.call_count == 1 |
| 113 | + |
| 114 | + # Metadata returned and stored (metadata flag being off doesn't matter because |
| 115 | + # grouping flag takes precedence) |
| 116 | + assert get_seer_similar_issues_return_values[0][0] == expected_metadata |
| 117 | + assert new_event.data["seer_similarity"] == expected_metadata |
| 118 | + |
| 119 | + # Parent group returned and used |
| 120 | + assert get_seer_similar_issues_return_values[0][1] == existing_event.group |
| 121 | + assert new_event.group_id == existing_event.group_id |
| 122 | + |
| 123 | + should_call_seer_spy.reset_mock() |
| 124 | + get_seer_similar_issues_spy.reset_mock() |
| 125 | + get_seer_similar_issues_return_values.pop() |
| 126 | + |
| 127 | + with Feature( |
| 128 | + { |
| 129 | + "projects:similarity-embeddings-metadata": True, |
| 130 | + "projects:similarity-embeddings-grouping": True, |
| 131 | + } |
| 132 | + ): |
| 133 | + new_event = save_new_event( |
| 134 | + {"message": "Cori and Bodhi are ridiculous"}, self.project |
| 135 | + ) |
| 136 | + expected_metadata = {**metadata_base, "request_hash": new_event.get_primary_hash()} |
| 137 | + |
| 138 | + # We checked whether to make the call, and then made it |
| 139 | + assert should_call_seer_spy.call_count == 1 |
| 140 | + assert get_seer_similar_issues_spy.call_count == 1 |
| 141 | + |
| 142 | + # Metadata returned and stored |
| 143 | + assert get_seer_similar_issues_return_values[0][0] == expected_metadata |
| 144 | + assert new_event.data["seer_similarity"] == expected_metadata |
| 145 | + |
| 146 | + # Parent group returned and used |
| 147 | + assert get_seer_similar_issues_return_values[0][1] == existing_event.group |
| 148 | + assert new_event.group_id == existing_event.group_id |
| 149 | + |
| 150 | + @with_feature("projects:similarity-embeddings-metadata") |
| 151 | + @patch("sentry.event_manager.get_seer_similar_issues", return_value=({}, None)) |
| 152 | + def test_calls_seer_if_no_group_found(self, mock_get_seer_similar_issues: MagicMock): |
| 153 | + save_new_event({"message": "Dogs are great!"}, self.project) |
| 154 | + assert mock_get_seer_similar_issues.call_count == 1 |
| 155 | + |
| 156 | + @with_feature("projects:similarity-embeddings-metadata") |
| 157 | + @patch("sentry.event_manager.get_seer_similar_issues", return_value=({}, None)) |
| 158 | + def test_bypasses_seer_if_group_found(self, mock_get_seer_similar_issues: MagicMock): |
| 159 | + existing_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 160 | + assert mock_get_seer_similar_issues.call_count == 1 |
| 161 | + |
| 162 | + new_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 163 | + assert existing_event.group_id == new_event.group_id |
| 164 | + assert mock_get_seer_similar_issues.call_count == 1 # didn't get called again |
| 165 | + |
| 166 | + @with_feature("projects:similarity-embeddings-metadata") |
| 167 | + def test_stores_seer_results_in_metadata(self): |
| 168 | + existing_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 169 | + |
| 170 | + seer_result_data = SeerSimilarIssueData( |
| 171 | + parent_hash=existing_event.get_primary_hash(), |
| 172 | + parent_group_id=NonNone(existing_event.group_id), |
| 173 | + stacktrace_distance=0.01, |
| 174 | + message_distance=0.05, |
| 175 | + should_group=True, |
| 176 | + ) |
| 177 | + |
| 178 | + with patch( |
| 179 | + "sentry.grouping.ingest.seer.get_similarity_data_from_seer", |
| 180 | + return_value=[seer_result_data], |
| 181 | + ): |
| 182 | + new_event = save_new_event({"message": "Adopt don't shop"}, self.project) |
| 183 | + expected_metadata = { |
| 184 | + "similarity_model_version": SEER_SIMILARITY_MODEL_VERSION, |
| 185 | + "request_hash": new_event.get_primary_hash(), |
| 186 | + "results": [asdict(seer_result_data)], |
| 187 | + } |
| 188 | + |
| 189 | + assert NonNone(new_event.group).data["metadata"]["seer_similarity"] == expected_metadata |
| 190 | + assert new_event.data["seer_similarity"] == expected_metadata |
| 191 | + |
| 192 | + @with_feature("projects:similarity-embeddings-grouping") |
| 193 | + def test_assigns_event_to_neighbor_group_if_found(self): |
| 194 | + existing_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 195 | + |
| 196 | + seer_result_data = SeerSimilarIssueData( |
| 197 | + parent_hash=existing_event.get_primary_hash(), |
| 198 | + parent_group_id=NonNone(existing_event.group_id), |
| 199 | + stacktrace_distance=0.01, |
| 200 | + message_distance=0.05, |
| 201 | + should_group=True, |
| 202 | + ) |
| 203 | + |
| 204 | + with patch( |
| 205 | + "sentry.grouping.ingest.seer.get_similarity_data_from_seer", |
| 206 | + return_value=[seer_result_data], |
| 207 | + ): |
| 208 | + new_event = save_new_event({"message": "Adopt don't shop"}, self.project) |
| 209 | + |
| 210 | + assert existing_event.group_id == new_event.group_id |
| 211 | + |
| 212 | + @with_feature("projects:similarity-embeddings-grouping") |
| 213 | + def test_creates_new_group_if_no_neighbor_found(self): |
| 214 | + existing_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 215 | + |
| 216 | + with patch("sentry.grouping.ingest.seer.get_similarity_data_from_seer", return_value=[]): |
| 217 | + new_event = save_new_event({"message": "Adopt don't shop"}, self.project) |
| 218 | + |
| 219 | + assert existing_event.group_id != new_event.group_id |
| 220 | + |
| 221 | + @with_feature("projects:similarity-embeddings-grouping") |
| 222 | + def test_creates_new_group_if_too_far_neighbor_found(self): |
| 223 | + existing_event = save_new_event({"message": "Dogs are great!"}, self.project) |
| 224 | + |
| 225 | + no_cigar_data = SeerSimilarIssueData( |
| 226 | + parent_hash=existing_event.get_primary_hash(), |
| 227 | + parent_group_id=NonNone(existing_event.group_id), |
| 228 | + stacktrace_distance=0.10, |
| 229 | + message_distance=0.05, |
| 230 | + should_group=False, |
| 231 | + ) |
| 232 | + |
| 233 | + with patch( |
| 234 | + "sentry.grouping.ingest.seer.get_similarity_data_from_seer", |
| 235 | + return_value=[no_cigar_data], |
| 236 | + ): |
| 237 | + new_event = save_new_event({"message": "Adopt don't shop"}, self.project) |
| 238 | + |
| 239 | + assert existing_event.group_id != new_event.group_id |
0 commit comments