Skip to content

:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.

feat(logs): enforce the logs alert cap on every create route (#129738)

Jimmy Zhu39b5b0349631
diff --git a/products/logs/backend/alert_creation.py b/products/logs/backend/alert_creation.py
index 5eb4ae9a6b82..c4413db79f66 100644
--- a/products/logs/backend/alert_creation.py
+++ b/products/logs/backend/alert_creation.py
@@ -4,10 +4,10 @@ import os
from datetime import UTC, datetime
from typing import Any
-from django.db import connection
-
from posthog.models.team.team import Team
+from products.alerts_platform.backend.facade.api import count_configurations, hold_cap_lock
+from products.alerts_platform.backend.facade.contracts import SourceKind
from products.logs.backend.alert_utils import next_allowed_check_at
from products.logs.backend.models import LogsAlertConfiguration
@@ -28,13 +28,22 @@ class AlertLimitReached(Exception):
def claim_alert_slot(team_id: int) -> None:
- """Serializes the team's creates until the caller's transaction ends, and raises
- `AlertLimitReached` when the team is at its cap."""
- with connection.cursor() as cursor:
- cursor.execute("SELECT pg_advisory_xact_lock(hashtext(%s))", [f"logs_alerts:{team_id}"])
+ """Serializes the team's logs alert creates on every route, the platform's included, until the
+ caller's transaction ends, and raises `AlertLimitReached` when the team is at its cap."""
+ hold_cap_lock(team_id=team_id, source_kind=SourceKind.LOGS)
+ check_alert_slot(team_id)
+
+
+def check_alert_slot(team_id: int) -> None:
+ """Raises `AlertLimitReached` when the team is at its cap. The caller holds the cap lock."""
if team_id in UNCAPPED_ALERT_TEAM_IDS:
return
- if LogsAlertConfiguration.objects.filter(team_id=team_id).count() >= MAX_ALERTS_PER_TEAM:
+ # Every logs row, and every platform alert with no logs row. A copy stands for a logs row already
+ # counted here, so copies are not counted.
+ alerts = LogsAlertConfiguration.objects.filter(team_id=team_id).count() + count_configurations(
+ team_id=team_id, source_kind=SourceKind.LOGS, native_only=True
+ )
+ if alerts >= MAX_ALERTS_PER_TEAM:
raise AlertLimitReached()
diff --git a/products/logs/backend/platform_ownership.py b/products/logs/backend/platform_ownership.py
index 6e72a1155007..c3c60fcaa10d 100644
--- a/products/logs/backend/platform_ownership.py
+++ b/products/logs/backend/platform_ownership.py
@@ -23,7 +23,7 @@ from products.alerts_platform.backend.facade.contracts import (
PlatformAlertCreateRefused,
PlatformAlertNotEditable,
)
-from products.logs.backend.alert_creation import AlertLimitReached, claim_alert_slot, first_check_at
+from products.logs.backend.alert_creation import AlertLimitReached, check_alert_slot, first_check_at
from products.logs.backend.alert_group_columns import grouping_problem
from products.logs.backend.alert_state_machine import apply_enabled
from products.logs.backend.facade.api import close_incident_before_delete
@@ -60,14 +60,27 @@ def on_created(alert: LogsAlertConfiguration, at: datetime, *, grouping: Groupin
return False
-def create_platform_owned(*, team_id: int, edit: PlatformAlertConfigurationEdit, created_by_id: int | None) -> UUID:
- """Creates a logs alert through the platform API: the logs row and the platform's copy, both owned
- by the platform.
- """
+def admit_platform_write(
+ *,
+ team_id: int,
+ current: PlatformAlertConfigurationEdit | None,
+ edit: PlatformAlertConfigurationEdit,
+ check_interval_minutes: int | None,
+) -> None:
+ """The per-team cap on a logs alert created through the platform. The platform holds the cap lock
+ the logs API's creates also take."""
+ if current is not None:
+ return
try:
- claim_alert_slot(team_id)
+ check_alert_slot(team_id)
except AlertLimitReached as error:
raise PlatformAlertCreateRefused(str(error)) from error
+
+
+def create_platform_owned(*, team_id: int, edit: PlatformAlertConfigurationEdit, created_by_id: int | None) -> UUID:
+ """Creates a logs alert through the platform API: the logs row and the platform's copy, both owned
+ by the platform. `admit_platform_write` has already checked the cap.
+ """
now = timezone.now()
alert = LogsAlertConfiguration(
team_id=team_id,
diff --git a/products/logs/backend/platform_source_config_schema.py b/products/logs/backend/platform_source_config_schema.py
index ff6e779c59e1..9040a46e69bf 100644
--- a/products/logs/backend/platform_source_config_schema.py
+++ b/products/logs/backend/platform_source_config_schema.py
@@ -20,7 +20,7 @@ from products.logs.backend.alert_deletion import delete_platform_owned
from products.logs.backend.alert_group_columns import grouping_problem
from products.logs.backend.alert_source_cycle import EVENT_IDS_BY_INCIDENT_ACTION
from products.logs.backend.models import LogsAlertConfiguration
-from products.logs.backend.platform_ownership import create_platform_owned, mirror_platform_edit
+from products.logs.backend.platform_ownership import admit_platform_write, create_platform_owned, mirror_platform_edit
from products.logs.backend.platform_source_config import WindowMinutes, has_any_filter, logs_alert_url
@@ -68,6 +68,7 @@ def validate_logs_edit(edit: PlatformAlertConfigurationEdit) -> None:
LOGS_SOURCE_CONFIG = SourceConfigSpec(
model=LogsPlatformAlertSourceConfig,
validate=validate_logs_edit,
+ admit=admit_platform_write,
mirror=mirror_platform_edit,
create=create_platform_owned,
delete=delete_platform_owned,
diff --git a/products/logs/backend/test/test_alerts_api.py b/products/logs/backend/test/test_alerts_api.py
index d06a0a1e9aac..ce47a85e0605 100644
--- a/products/logs/backend/test/test_alerts_api.py
+++ b/products/logs/backend/test/test_alerts_api.py
@@ -726,19 +726,36 @@ class TestLogsAlertAPI(APIBaseTest):
@parameterized.expand(
[
- ("capped", False, status.HTTP_400_BAD_REQUEST, MAX_ALERTS_PER_TEAM),
- ("uncapped", True, status.HTTP_201_CREATED, MAX_ALERTS_PER_TEAM + 1),
+ ("capped", False, "logs_rows", status.HTTP_400_BAD_REQUEST, 0),
+ ("uncapped", True, "logs_rows", status.HTTP_201_CREATED, 1),
+ ("capped_by_alerts_only_on_the_platform", False, "platform_only", status.HTTP_400_BAD_REQUEST, 0),
+ ("not_capped_by_copies_of_deleted_alerts", False, "copies_of_deleted_rows", status.HTTP_201_CREATED, 1),
]
)
- def test_per_team_limit(self, _name, uncapped, expected_status, expected_count):
+ def test_per_team_limit(
+ self, _name: str, uncapped: bool, existing: str, expected_status: int, created: int
+ ) -> None:
for i in range(MAX_ALERTS_PER_TEAM):
- LogsAlertConfiguration.objects.create(
- team=self.team,
- name=f"Alert {i}",
- threshold_count=1,
- created_by=self.user,
- filters={"severityLevels": ["error"]},
- )
+ if existing != "logs_rows":
+ with team_scope(self.team.id):
+ platform_testing.create_configuration(
+ team_id=self.team.id,
+ name=f"Alert {i}",
+ source_kind="logs",
+ source_config={},
+ check_interval_minutes=5,
+ owner="platform" if existing == "platform_only" else "source",
+ legacy_configuration_id=uuid4() if existing == "copies_of_deleted_rows" else None,
+ )
+ else:
+ LogsAlertConfiguration.objects.create(
+ team=self.team,
+ name=f"Alert {i}",
+ threshold_count=1,
+ created_by=self.user,
+ filters={"severityLevels": ["error"]},
+ )
+ logs_rows_before = LogsAlertConfiguration.objects.filter(team=self.team).count()
patched_ids = frozenset({self.team.id}) if uncapped else frozenset()
with patch("products.logs.backend.alert_creation.UNCAPPED_ALERT_TEAM_IDS", patched_ids):
@@ -748,7 +765,7 @@ class TestLogsAlertAPI(APIBaseTest):
format="json",
)
assert response.status_code == expected_status
- assert LogsAlertConfiguration.objects.filter(team=self.team).count() == expected_count
+ assert LogsAlertConfiguration.objects.filter(team=self.team).count() == logs_rows_before + created
# --- Read-only fields ---