ref(seer-grouping): v2.1 flag (#112841)

kddubey · web-flow · commit fe3ba9301647 · 2026-04-14T14:08:34.000-07:00
temp flag to migrate EA projects from v2 to v2.1 can be merged independently of [seer PR stack](getsentry/seer#5735) while no project is flagged in
diff --git a/src/sentry/features/temporary.py b/src/sentry/features/temporary.py
@@ -570,6 +570,8 @@ def register_temporary_features(manager: FeatureManager) -> None:
     manager.add("projects:similarity-view", ProjectFeature, FeatureHandlerStrategy.INTERNAL, api_expose=True)
     # Enable new similarity grouping model upgrade (version-agnostic rollout)
     manager.add("projects:similarity-grouping-model-upgrade", ProjectFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=False)
+    # Enable next similarity grouping model rollout (for migrating EA projects from v2 to v2.1)
+    manager.add("projects:similarity-grouping-model-next", ProjectFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=False)
     # Starfish: extract metrics from the spans
     manager.add("projects:span-metrics-extraction", ProjectFeature, FeatureHandlerStrategy.INTERNAL, api_expose=True)
     manager.add("projects:span-metrics-extraction-addons", ProjectFeature, FeatureHandlerStrategy.INTERNAL, api_expose=False)
diff --git a/src/sentry/grouping/ingest/seer.py b/src/sentry/grouping/ingest/seer.py
@@ -19,7 +19,6 @@
 from sentry.seer.signed_seer_api import SeerViewerContext
 from sentry.seer.similarity.config import (
     get_grouping_model_version,
-    get_new_model_version,
     should_send_to_seer_for_training,
 )
 from sentry.seer.similarity.similar_issues import get_similarity_data_from_seer
@@ -624,10 +623,10 @@ def maybe_send_seer_for_new_model_training(
     variants: dict[str, BaseVariant],
 ) -> None:
     """
-    Send a training_mode=true request to Seer for the new model version if the existing
-    grouphash hasn't been sent to the new version yet.
+    Send a training_mode=true request to Seer for the project's current non-stable model
+    version if the existing grouphash hasn't been sent to that version yet.
 
-    This only happens for projects that have the new model rolled out. It helps
+    This only happens for projects on a non-stable model (via feature flags). It helps
     build data for existing groups without affecting production grouping decisions.
 
     Args:
@@ -688,10 +687,10 @@ def maybe_send_seer_for_new_model_training(
         },
     )
 
-    # Mark the grouphash as sent to the new model so we don't send duplicate requests.
+    # Mark the grouphash as sent to this (non-stable) model so we don't send duplicate requests.
     # We update seer_latest_training_model (not seer_model) to preserve the original
     # grouping decision metadata.
     if gh_metadata:
-        new_version = get_new_model_version()
-        if new_version is not None:
-            gh_metadata.update(seer_latest_training_model=new_version.value)
+        gh_metadata.update(
+            seer_latest_training_model=get_grouping_model_version(event.project).value
+        )
diff --git a/src/sentry/seer/similarity/config.py b/src/sentry/seer/similarity/config.py
@@ -18,78 +18,53 @@
 # Set to None to disable rollout entirely
 SEER_GROUPING_NEW_VERSION: GroupingVersion | None = GroupingVersion.V2
 
-# Feature flag name (version-agnostic)
+# Model version to migrate EA projects from new to next
+SEER_GROUPING_NEXT_VERSION: GroupingVersion | None = GroupingVersion.V2_1
+
+# Feature flag names
 SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE = "projects:similarity-grouping-model-upgrade"
+SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE = "projects:similarity-grouping-model-next"
 
 
 def get_grouping_model_version(project: Project) -> GroupingVersion:
     """
     Get the model version to use for grouping decisions for this project.
 
     Returns:
+        - Next version if rollout is enabled for this project (for migrating EA projects from new to next)
         - New version if rollout is enabled for this project
         - Stable version otherwise
     """
-    # Early return if no new version configured
-    if SEER_GROUPING_NEW_VERSION is None:
-        return SEER_GROUPING_STABLE_VERSION
-
-    # Type is narrowed to GroupingVersion here
-    if features.has(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, project):
+    if SEER_GROUPING_NEXT_VERSION is not None and features.has(
+        SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE, project
+    ):
+        return SEER_GROUPING_NEXT_VERSION
+
+    if SEER_GROUPING_NEW_VERSION is not None and features.has(
+        SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, project
+    ):
         return SEER_GROUPING_NEW_VERSION
-    return SEER_GROUPING_STABLE_VERSION
-
-
-def is_new_model_rolled_out(project: Project) -> bool:
-    """
-    Check if the new model version is rolled out for this project.
-
-    Returns False if:
-    - No new version is configured (rollout disabled globally)
-    - Feature flag is not enabled for this project
-    """
-    if SEER_GROUPING_NEW_VERSION is None:
-        return False
-
-    return features.has(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, project)
 
-
-def get_new_model_version() -> GroupingVersion | None:
-    """
-    Get the new model version being rolled out, if any.
-    Returns None if no rollout is in progress.
-    """
-    return SEER_GROUPING_NEW_VERSION
+    return SEER_GROUPING_STABLE_VERSION
 
 
 def should_send_to_seer_for_training(
     project: Project,
     grouphash_seer_latest_training_model: str | None,
 ) -> bool:
     """
-    Check if we should send a training_mode=true request to Seer for the new model version.
+    Check if we should send a training_mode=true request to Seer for the
+    project's current model version.
 
     This is true when:
-    1. A new version is being rolled out
-    2. The project has the rollout feature enabled
-    3. The grouphash hasn't already been sent to the new version
-
-    Args:
-        project: The project
-        grouphash_seer_latest_training_model: The seer_latest_training_model value
-            from grouphash metadata
-
-    Returns:
-        True if we should send a training_mode=true request
+    1. The project is on a non-stable model version (via feature flags)
+    2. The grouphash hasn't already been sent to that version
     """
-    new_version = get_new_model_version()
-    if new_version is None:
-        return False
-
-    if not is_new_model_rolled_out(project):
+    model_version = get_grouping_model_version(project)
+    if model_version == SEER_GROUPING_STABLE_VERSION:
         return False
 
-    if grouphash_seer_latest_training_model == new_version.value:
+    if grouphash_seer_latest_training_model == model_version.value:
         return False
 
     return True
diff --git a/src/sentry/seer/similarity/types.py b/src/sentry/seer/similarity/types.py
@@ -14,6 +14,7 @@ class GroupingVersion(StrEnum):
 
     V1 = "v1"
     V2 = "v2"
+    V2_1 = "v2.1"
 
 
 class IncompleteSeerDataError(Exception):
diff --git a/src/sentry/seer/similarity/utils.py b/src/sentry/seer/similarity/utils.py
@@ -347,13 +347,13 @@ def stacktrace_exceeds_limits(
     event: Event | GroupEvent,
     variants: dict[str, BaseVariant],
     referrer: ReferrerOptions,
-    model_version: GroupingVersion | None = None,
+    model_version: GroupingVersion,
 ) -> bool:
     """
     Check if a stacktrace exceeds length limits for Seer similarity analysis.
 
     For V1, platforms that bypass length checks (to maintain consistency with backfilled data)
-    have all stacktraces pass through. For V2, all platforms are subject to length checks.
+    have all stacktraces pass through. For non-V1 models, all platforms are subject to length checks.
 
     If we dont bypass length checks, we use a two-step approach:
     1. First check raw string length - if shorter than token limit, pass immediately
@@ -381,7 +381,7 @@ def stacktrace_exceeds_limits(
     # matching with existing data, we bypass the filter for them (their stacktraces will be truncated).
     # For V2 we apply length checks to all platforms since we're re-embedding everything anyway.
     if (
-        model_version != GroupingVersion.V2
+        model_version == GroupingVersion.V1
         and platform in EVENT_PLATFORMS_BYPASSING_STACKTRACE_LENGTH_CHECK
     ):
         metrics.incr(
diff --git a/tests/sentry/grouping/seer_similarity/test_training_mode.py b/tests/sentry/grouping/seer_similarity/test_training_mode.py
@@ -3,7 +3,10 @@
 from sentry.grouping.ingest.seer import maybe_send_seer_for_new_model_training
 from sentry.models.grouphash import GroupHash
 from sentry.models.grouphashmetadata import GroupHashMetadata
-from sentry.seer.similarity.config import SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE
+from sentry.seer.similarity.config import (
+    SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE,
+    SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE,
+)
 from sentry.testutils.cases import TestCase
 from sentry.testutils.helpers.eventprocessing import save_new_event
 
@@ -30,6 +33,7 @@ def test_does_nothing_when_no_rollout(self) -> None:
         """Should not send request when no new version is being rolled out"""
         with (
             patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None),
+            patch("sentry.seer.similarity.config.SEER_GROUPING_NEXT_VERSION", None),
             patch(
                 "sentry.grouping.ingest.seer.get_similarity_data_from_seer"
             ) as mock_get_similarity_data,
@@ -212,3 +216,26 @@ def test_captures_exception_without_failing(self) -> None:
                     "grouphash": self.grouphash.hash,
                 },
             )
+
+    def test_retrains_for_next_model_when_already_trained_for_new(self) -> None:
+        """v2 -> v2.1 transition: grouphash trained for v2, project now on v2.1"""
+        with (
+            patch("sentry.grouping.ingest.seer.should_call_seer_for_grouping", return_value=True),
+            patch(
+                "sentry.grouping.ingest.seer.get_similarity_data_from_seer",
+                return_value=([], "v2.1"),
+            ) as mock_get_similarity_data,
+            self.feature(SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE),
+        ):
+            metadata, _ = GroupHashMetadata.objects.get_or_create(grouphash=self.grouphash)
+            metadata.seer_latest_training_model = "v2"
+            metadata.save()
+
+            maybe_send_seer_for_new_model_training(self.event, self.grouphash, self.variants)
+
+            mock_get_similarity_data.assert_called_once()
+            call_args = mock_get_similarity_data.call_args
+            assert call_args[0][0]["model"].value == "v2.1"
+
+            metadata.refresh_from_db()
+            assert metadata.seer_latest_training_model == "v2.1"
diff --git a/tests/sentry/seer/similarity/test_config.py b/tests/sentry/seer/similarity/test_config.py
@@ -3,103 +3,95 @@
 from sentry.seer.similarity.config import (
     SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE,
     SEER_GROUPING_NEW_VERSION,
+    SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE,
+    SEER_GROUPING_NEXT_VERSION,
     SEER_GROUPING_STABLE_VERSION,
     get_grouping_model_version,
-    get_new_model_version,
-    is_new_model_rolled_out,
     should_send_to_seer_for_training,
 )
-from sentry.seer.similarity.types import GroupingVersion
 from sentry.testutils.cases import TestCase
 
 
 class GetGroupingModelVersionTest(TestCase):
-    def test_returns_stable_when_rollout_disabled(self) -> None:
-        """When new model rollout is disabled, return stable version"""
-        with patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None):
-            result = get_grouping_model_version(self.project)
-            assert result == SEER_GROUPING_STABLE_VERSION
-
-    def test_returns_stable_when_feature_not_enabled(self) -> None:
-        """When feature flag is not enabled for project, return stable version"""
-        result = get_grouping_model_version(self.project)
-        assert result == SEER_GROUPING_STABLE_VERSION
-
-    def test_returns_new_when_feature_enabled(self) -> None:
-        """When feature flag is enabled for project, return new version"""
-        with self.feature(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE):
-            result = get_grouping_model_version(self.project)
-            assert result == SEER_GROUPING_NEW_VERSION
-
-
-class IsNewModelRolledOutTest(TestCase):
-    def test_returns_false_when_no_new_version(self) -> None:
-        """When no new version is configured, rollout is not active"""
-        with patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None):
-            result = is_new_model_rolled_out(self.project)
-            assert result is False
-
-    def test_returns_false_when_feature_not_enabled(self) -> None:
-        """When feature flag is not enabled, rollout is not active for project"""
-        result = is_new_model_rolled_out(self.project)
-        assert result is False
-
-    def test_returns_true_when_feature_enabled(self) -> None:
-        """When feature flag is enabled, rollout is active for project"""
-        with self.feature(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE):
-            result = is_new_model_rolled_out(self.project)
-            assert result is True
-
-
-class GetNewModelVersionTest(TestCase):
-    def test_returns_configured_version(self) -> None:
-        """Returns the configured new model version"""
-        result = get_new_model_version()
-        assert result == GroupingVersion.V2
+    def test_returns_stable_when_no_flags(self) -> None:
+        assert get_grouping_model_version(self.project) == SEER_GROUPING_STABLE_VERSION
 
-    def test_returns_none_when_disabled(self) -> None:
-        """Returns None when rollout is disabled"""
-        with patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None):
-            result = get_new_model_version()
-            assert result is None
+    def test_returns_stable_when_rollout_disabled(self) -> None:
+        with (
+            patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None),
+            patch("sentry.seer.similarity.config.SEER_GROUPING_NEXT_VERSION", None),
+        ):
+            assert get_grouping_model_version(self.project) == SEER_GROUPING_STABLE_VERSION
+
+    def test_returns_flagged_version(self) -> None:
+        cases = [
+            (SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, SEER_GROUPING_NEW_VERSION),
+            (SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE, SEER_GROUPING_NEXT_VERSION),
+        ]
+        for feature_flag, expected_version in cases:
+            with self.subTest(feature_flag=feature_flag), self.feature(feature_flag):
+                assert get_grouping_model_version(self.project) == expected_version
+
+    def test_next_flag_takes_priority_over_new_flag(self) -> None:
+        with self.feature(
+            [SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE]
+        ):
+            assert get_grouping_model_version(self.project) == SEER_GROUPING_NEXT_VERSION
+
+    def test_falls_back_to_new_when_next_version_is_none(self) -> None:
+        with (
+            patch("sentry.seer.similarity.config.SEER_GROUPING_NEXT_VERSION", None),
+            self.feature(
+                [SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE]
+            ),
+        ):
+            assert get_grouping_model_version(self.project) == SEER_GROUPING_NEW_VERSION
 
 
 class ShouldSendToSeerForTrainingTest(TestCase):
     def test_returns_false_when_no_rollout(self) -> None:
-        """Returns False when no new version is being rolled out"""
-        with patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None):
+        with (
+            patch("sentry.seer.similarity.config.SEER_GROUPING_NEW_VERSION", None),
+            patch("sentry.seer.similarity.config.SEER_GROUPING_NEXT_VERSION", None),
+        ):
             result = should_send_to_seer_for_training(
                 self.project, grouphash_seer_latest_training_model=None
             )
             assert result is False
 
-    def test_returns_false_when_feature_not_enabled(self) -> None:
-        """Returns False when feature flag is not enabled for project"""
+    def test_returns_false_when_no_flags(self) -> None:
         result = should_send_to_seer_for_training(
             self.project, grouphash_seer_latest_training_model=None
         )
         assert result is False
 
-    def test_returns_true_when_never_sent(self) -> None:
-        """Returns True when grouphash has never been sent to Seer"""
-        with self.feature(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE):
-            result = should_send_to_seer_for_training(
-                self.project, grouphash_seer_latest_training_model=None
-            )
-            assert result is True
-
-    def test_returns_true_when_sent_to_old_version(self) -> None:
-        """Returns True when grouphash was sent to an older model version"""
-        with self.feature(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE):
-            result = should_send_to_seer_for_training(
-                self.project, grouphash_seer_latest_training_model="v1"
-            )
-            assert result is True
-
-    def test_returns_false_when_already_sent_to_new_version(self) -> None:
-        """Returns False when grouphash was already sent to the new version"""
-        with self.feature(SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE):
-            result = should_send_to_seer_for_training(
-                self.project, grouphash_seer_latest_training_model="v2"
-            )
-            assert result is False
+    def test_returns_true_when_training_needed(self) -> None:
+        cases = [
+            # (feature_flag, seer_latest_training_model)
+            (SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, None),
+            (SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, "v1"),
+            (SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE, None),
+            (SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE, "v1"),
+            (SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE, "v2"),  # v2 → v2.1 transition
+        ]
+        for feature_flag, training_model in cases:
+            with self.subTest(feature_flag=feature_flag, training_model=training_model):
+                with self.feature(feature_flag):
+                    result = should_send_to_seer_for_training(
+                        self.project, grouphash_seer_latest_training_model=training_model
+                    )
+                    assert result is True
+
+    def test_returns_false_when_already_sent_to_current_version(self) -> None:
+        cases = [
+            # (feature_flag, seer_latest_training_model)
+            (SEER_GROUPING_NEW_MODEL_ROLLOUT_FEATURE, "v2"),
+            (SEER_GROUPING_NEXT_MODEL_ROLLOUT_FEATURE, "v2.1"),
+        ]
+        for feature_flag, training_model in cases:
+            with self.subTest(feature_flag=feature_flag, training_model=training_model):
+                with self.feature(feature_flag):
+                    result = should_send_to_seer_for_training(
+                        self.project, grouphash_seer_latest_training_model=training_model
+                    )
+                    assert result is False
diff --git a/tests/sentry/seer/similarity/test_utils.py b/tests/sentry/seer/similarity/test_utils.py
