Merge branch 'master' into kcons/filtered

Author: kcons

migrations_lockfile.txt

@@ -21,7 +21,7 @@ remote_subscriptions: 0003_drop_remote_subscription
 
 replays: 0005_drop_replay_index
 
-sentry: 0894_split_discover_dataset_saved_queries
+sentry: 0895_relocation_provenance_smallint
 
 social_auth: 0002_default_auto_field
 

src/sentry/grouping/grouptype.py

@@ -5,9 +5,13 @@
 from sentry.models.group import DEFAULT_TYPE_ID
 from sentry.types.group import PriorityLevel
 from sentry.workflow_engine.endpoints.validators.error_detector import ErrorDetectorValidator
-from sentry.workflow_engine.handlers.detector.base import DetectorEvaluationResult, DetectorHandler
+from sentry.workflow_engine.handlers.detector.base import DetectorHandler
 from sentry.workflow_engine.models.data_source import DataPacket
-from sentry.workflow_engine.types import DetectorGroupKey, DetectorSettings
+from sentry.workflow_engine.types import (
+    DetectorEvaluationResult,
+    DetectorGroupKey,
+    DetectorSettings,
+)
 
 T = TypeVar("T")
 

src/sentry/incidents/grouptype.py

@@ -18,6 +18,7 @@
 )
 from sentry.workflow_engine.handlers.detector.base import EvidenceData
 from sentry.workflow_engine.models.data_source import DataPacket
+from sentry.workflow_engine.processors.data_condition_group import ProcessedDataConditionGroup
 from sentry.workflow_engine.types import DetectorGroupKey, DetectorPriorityLevel, DetectorSettings
 
 COMPARISON_DELTA_CHOICES: list[None | int] = [choice.value for choice in ComparisonDeltaChoices]
@@ -32,7 +33,8 @@ class MetricIssueEvidenceData(EvidenceData):
 class MetricAlertDetectorHandler(StatefulGroupingDetectorHandler[QuerySubscriptionUpdate, int]):
     def create_occurrence(
         self,
-        value: int,
+        evaluation_result: ProcessedDataConditionGroup,
+        data_packet: DataPacket[QuerySubscriptionUpdate],
         priority: DetectorPriorityLevel,
     ) -> tuple[DetectorOccurrence, dict[str, Any]]:
         # Returning a placeholder for now, this may require us passing more info

src/sentry/migrations/0895_relocation_provenance_smallint.py

@@ -0,0 +1,35 @@
+# Generated by Django 5.2.1 on 2025-05-14 18:28
+
+from sentry.new_migrations.migrations import CheckedMigration
+from sentry.new_migrations.monkey.special import SafeRunSQL
+
+
+class Migration(CheckedMigration):
+    # This flag is used to mark that a migration shouldn't be automatically run in production.
+    # This should only be used for operations where it's safe to run the migration after your
+    # code has deployed. So this should not be used for most operations that alter the schema
+    # of a table.
+    # Here are some things that make sense to mark as post deployment:
+    # - Large data migrations. Typically we want these to be run manually so that they can be
+    #   monitored and not block the deploy for a long period of time while they run.
+    # - Adding indexes to large tables. Since this can take a long time, we'd generally prefer to
+    #   run this outside deployments so that we don't block them. Note that while adding an index
+    #   is a schema change, it's completely safe to run the operation after the code has deployed.
+    # Once deployed, run these manually via: https://develop.sentry.dev/database-migrations/#migration-deployment
+
+    is_post_deployment = True
+
+    dependencies = [
+        ("sentry", "0894_split_discover_dataset_saved_queries"),
+    ]
+
+    operations = [
+        SafeRunSQL(
+            """
+            ALTER TABLE sentry_relocation
+            ALTER COLUMN provenance TYPE smallint;
+            """,
+            reverse_sql="",
+            hints={"tables": ["sentry_relocation"]},
+        ),
+    ]

src/sentry/monitors/consumers/monitor_consumer.py

@@ -6,7 +6,7 @@
 from collections.abc import Mapping
 from concurrent.futures import ThreadPoolExecutor, wait
 from copy import deepcopy
-from datetime import UTC, datetime, timedelta
+from datetime import UTC, datetime
 from functools import partial
 from typing import Any, Literal, NotRequired, TypedDict
 
@@ -811,20 +811,14 @@ def _process_checkin(item: CheckinItem, txn: Transaction | Span) -> None:
             # 03-B
             # Create a brand new check-in object
             except MonitorCheckIn.DoesNotExist:
-                # Infer the original start time of the check-in from the duration.
-                # Note that the clock of this worker may be off from what Relay is reporting.
-                date_added = start_time
-                if duration is not None:
-                    date_added -= timedelta(milliseconds=duration)
-
                 # When was this check-in expected to have happened?
                 expected_time = monitor_environment.next_checkin
 
                 # denormalize the monitor configration into the check-in.
                 # Useful to show details about the configuration of the
                 # monitor at the time of the check-in
                 monitor_config = monitor.get_validated_config()
-                timeout_at = get_timeout_at(monitor_config, status, date_added)
+                timeout_at = get_timeout_at(monitor_config, status, start_time)
 
                 # The "date_clock" is recorded as the "clock time" of when the
                 # check-in was processed. The clock time is derived from the
@@ -839,9 +833,9 @@ def _process_checkin(item: CheckinItem, txn: Transaction | Span) -> None:
                     defaults={
                         "duration": duration,
                         "status": status,
-                        "date_added": date_added,
-                        "date_clock": clock_time,
+                        "date_added": start_time,
                         "date_updated": start_time,
+                        "date_clock": clock_time,
                         "expected_time": expected_time,
                         "timeout_at": timeout_at,
                         "monitor_config": monitor_config,

src/sentry/monitors/models.py

@@ -465,14 +465,25 @@ class MonitorCheckIn(Model):
     check-in.
     """
 
+    date_created = models.DateTimeField(default=timezone.now, null=True)
+    """
+    Represents when the check-in was actually recorded into the database. This
+    is a real wall-clock time and is not tied to the "clock" time that
+    check-ins are processed in the context of.
+    """
+
     date_added = models.DateTimeField(default=timezone.now, db_index=True)
     """
-    Represents the time the checkin was made. This CAN BE back-dated in some
-    cases, and does not necessarily represent the insertion time of the row in
-    the database.
+    Represents the time the checkin was made. This date comes from the time
+    relay received the envelope containing the check-in.
+    """
 
-    This date comes from the time relay reiceved the envelope containing the
-    check-in.
+    date_updated = models.DateTimeField(default=timezone.now)
+    """
+    Represents the last time a check-in was updated by . This will typically be by the terminal state
+    Currently only updated when a in_progress check-in is sent with this
+    check-in's guid. Can be used to extend the lifetime of a check-in so that
+    it does not time out.
     """
 
     date_clock = models.DateTimeField(null=True)
@@ -484,20 +495,6 @@ class MonitorCheckIn(Model):
     as detecting misses)
     """
 
-    date_created = models.DateTimeField(default=timezone.now, null=True)
-    """
-    Represents when the check-in was actually recorded into the database. This
-    is a real wall-clock time and is not tied to the "clock" time that
-    check-ins are processed in the contenxt of.
-    """
-
-    date_updated = models.DateTimeField(default=timezone.now)
-    """
-    Currently only updated when a in_progress check-in is sent with this
-    check-in's guid. Can be used to extend the lifetime of a check-in so that
-    it does not time out.
-    """
-
     expected_time = models.DateTimeField(null=True)
     """
     Holds the exact time we expected to receive this check-in

src/sentry/monitors/serializers.py

@@ -275,7 +275,10 @@ class MonitorCheckInSerializerResponse(MonitorCheckInSerializerResponseOptional)
     environment: str
     status: str
     duration: int
+    # TODO(epurkhiser): Should be replaced with the actual date_created, right
+    # now this is the same as dateAdded
     dateCreated: datetime
+    dateAdded: datetime
     expectedTime: datetime
     monitorConfig: MonitorConfigSerializerResponse
 
@@ -342,6 +345,7 @@ def serialize(self, obj, attrs, user, **kwargs) -> MonitorCheckInSerializerRespo
             "status": obj.get_status_display(),
             "duration": obj.duration,
             "dateCreated": obj.date_added,
+            "dateAdded": obj.date_added,
             "expectedTime": obj.expected_time,
             "monitorConfig": cast(MonitorConfigSerializerResponse, config),
         }

src/sentry/rules/processing/buffer_processing.py

@@ -104,6 +104,7 @@ def process_in_batches(project_id: int, processing_type: str) -> None:
     metrics.incr(
         f"{processing_type}.num_groups", tags={"num_groups": bucket_num_groups(event_count)}
     )
+    metrics.distribution(f"{processing_type}.event_count", event_count)
 
     if event_count < batch_size:
         return task.delay(project_id)

src/sentry/workflow_engine/handlers/detector/__init__.py

@@ -1,18 +1,11 @@
 __all__ = [
     "DataPacketEvaluationType",
     "DataPacketType",
-    "DetectorEvaluationResult",
     "DetectorHandler",
     "DetectorOccurrence",
     "DetectorStateData",
     "StatefulGroupingDetectorHandler",
 ]
 
-from .base import (
-    DataPacketEvaluationType,
-    DataPacketType,
-    DetectorEvaluationResult,
-    DetectorHandler,
-    DetectorOccurrence,
-)
+from .base import DataPacketEvaluationType, DataPacketType, DetectorHandler, DetectorOccurrence
 from .stateful import DetectorStateData, StatefulGroupingDetectorHandler

src/sentry/workflow_engine/handlers/detector/base.py

@@ -8,15 +8,21 @@
 
 from sentry.issues.grouptype import GroupType
 from sentry.issues.issue_occurrence import IssueEvidence, IssueOccurrence
-from sentry.issues.status_change_message import StatusChangeMessage
 from sentry.types.actor import Actor
 from sentry.workflow_engine.models import Condition, DataConditionGroup, DataPacket, Detector
-from sentry.workflow_engine.types import DetectorGroupKey, DetectorPriorityLevel
+from sentry.workflow_engine.processors.data_condition_group import ProcessedDataConditionGroup
+from sentry.workflow_engine.types import (
+    DetectorEvaluationResult,
+    DetectorGroupKey,
+    DetectorPriorityLevel,
+)
 
 logger = logging.getLogger(__name__)
 
 DataPacketType = TypeVar("DataPacketType")
 DataPacketEvaluationType = TypeVar("DataPacketEvaluationType")
+
+# TODO - get more info about how this is used in issue platform
 EventData = dict[str, Any]
 
 
@@ -35,14 +41,14 @@ class EvidenceData(Generic[DataPacketEvaluationType]):
 class DetectorOccurrence:
     issue_title: str
     subtitle: str
-    resource_id: str | None = None
     evidence_data: Mapping[str, Any] = dataclasses.field(default_factory=dict)
     evidence_display: Sequence[IssueEvidence] = dataclasses.field(default_factory=list)
     type: type[GroupType]
     level: str
     culprit: str
-    priority: int | None = None
+    resource_id: str | None = None
     assignee: Actor | None = None
+    priority: DetectorPriorityLevel | None = None
 
     def to_issue_occurrence(
         self,
@@ -73,20 +79,6 @@ def to_issue_occurrence(
         )
 
 
-@dataclasses.dataclass(frozen=True)
-class DetectorEvaluationResult:
-    # TODO - Should group key live at this level?
-    group_key: DetectorGroupKey
-    # TODO: Are these actually necessary? We're going to produce the occurrence in the detector, so we probably don't
-    # need to know the other results externally
-    is_triggered: bool
-    priority: DetectorPriorityLevel
-    # TODO: This is only temporarily optional. We should always have a value here if returning a result
-    result: IssueOccurrence | StatusChangeMessage | None = None
-    # Event data to supplement the `IssueOccurrence`, if passed.
-    event_data: dict[str, Any] | None = None
-
-
 # TODO - DetectorHandler -> AbstractDetectorHandler? (then DetectorHandler is the base implementation)
 class DetectorHandler(abc.ABC, Generic[DataPacketType, DataPacketEvaluationType]):
     def __init__(self, detector: Detector):
@@ -109,7 +101,7 @@ def __init__(self, detector: Detector):
     @abc.abstractmethod
     def evaluate(
         self, data_packet: DataPacket[DataPacketType]
-    ) -> dict[DetectorGroupKey, DetectorEvaluationResult]:
+    ) -> dict[DetectorGroupKey, DetectorEvaluationResult] | None:
         """
         This method is used to evaluate the data packet's value against the conditions on the detector.
         """
@@ -118,10 +110,9 @@ def evaluate(
     @abc.abstractmethod
     def create_occurrence(
         self,
-        value: DataPacketEvaluationType,
+        evaluation_result: ProcessedDataConditionGroup,
+        data_packet: DataPacket[DataPacketType],
         priority: DetectorPriorityLevel,
-        # data_packet: DataPacketType, # TODO - having access to all the data being evaluated seems good
-        # data_conditions: list[DataCondition], # TODO - list of the failing conditions might be nice
     ) -> tuple[DetectorOccurrence, EventData]:
         """
         This method provides the value that was evaluated against, the data packet that was