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.

perf(signals): sort the scout findings summary without a task run join (#120867)

posthog[bot]b36188859f5d
diff --git a/products/signals/backend/scout_harness/tools/runs.py b/products/signals/backend/scout_harness/tools/runs.py
index 4b93a083cf98..528ab9f54571 100644
--- a/products/signals/backend/scout_harness/tools/runs.py
+++ b/products/signals/backend/scout_harness/tools/runs.py
@@ -26,7 +26,6 @@ from uuid import UUID
from django.db import connection
from django.db.models import Q
-from django.db.models.functions import Coalesce
from django.utils import timezone
import structlog
@@ -536,26 +535,39 @@ def fleet_findings_summary(*, team_id: int, window_hours: int = DEFAULT_FINDINGS
touched_a_report = (~Q(emitted_report_ids=[]) & ~Q(emitted_report_ids__isnull=True)) | (
~Q(edited_report_ids=[]) & ~Q(edited_report_ids__isnull=True)
)
- # Load the ids first. As an `id__in` subquery, the planner can start from `posthog_task_run`,
- # which holds every product's task runs, instead of from these few scout runs.
- candidate_ids = list(
+ # No index can serve a sort on `Coalesce(task_run.completed_at, created_at)`, so read the
+ # candidates without a task run join and sort them below on completion times read by primary key.
+ candidates = list(
SignalScoutRun.objects.filter(team_id=team_id, created_at__gte=window_start)
.exclude(metadata__has_key="scout_trial")
.filter(Q(emitted_count__gt=0) | touched_a_report)
.order_by("-created_at", "-id")
- .values_list("id", flat=True)[:FLEET_FINDINGS_SUMMARY_CANDIDATE_RUNS]
+ .values_list(
+ "id",
+ "task_run_id",
+ "created_at",
+ "emitted_count",
+ "skill_name",
+ "emitted_report_ids",
+ "edited_report_ids",
+ )[:FLEET_FINDINGS_SUMMARY_CANDIDATE_RUNS]
)
+ completed_at_by_task_run = tasks_facade.get_task_run_completed_at(
+ [candidate[1] for candidate in candidates], team_id=team_id
+ )
+
+ def run_emitted_at(task_run_id: UUID, created_at: datetime) -> datetime:
+ return completed_at_by_task_run.get(task_run_id) or created_at
+
# Order by completion (fall back to creation) so the cap keeps the *most recently emitted* runs,
# matching the frontend's `completed_at ?? created_at` sort; `-id` tie-breaks on the time-ordered PK.
- rows = (
- SignalScoutRun.objects.filter(team_id=team_id, id__in=candidate_ids)
- .annotate(_emitted_at=Coalesce("task_run__completed_at", "created_at"))
- .order_by("-_emitted_at", "-id")
- .values_list("emitted_count", "skill_name", "_emitted_at", "emitted_report_ids", "edited_report_ids")[
+ ranked = sorted(candidates, key=lambda c: (run_emitted_at(c[1], c[2]), c[0]), reverse=True)
+ materialized = [
+ (emitted_count, skill_name, run_emitted_at(task_run_id, created_at), emitted_report_ids, edited_report_ids)
+ for _, task_run_id, created_at, emitted_count, skill_name, emitted_report_ids, edited_report_ids in ranked[
:FLEET_FINDINGS_SUMMARY_RUN_CAP
]
- )
- materialized = list(rows)
+ ]
count = 0
scouts: set[str] = set()
latest_at: datetime | None = None
diff --git a/products/tasks/backend/facade/api.py b/products/tasks/backend/facade/api.py
index 8e16fbd9b38c..2d4c078c41cd 100644
--- a/products/tasks/backend/facade/api.py
+++ b/products/tasks/backend/facade/api.py
@@ -303,6 +303,7 @@ __all__ = [
"task_access_scope",
"normalize_verified_pull_request_urls_batch",
"get_task_id_for_run",
+ "get_task_run_completed_at",
"get_task_run",
"get_task_run_session",
"sync_task_run_session",
@@ -1198,6 +1199,17 @@ def get_task_id_for_run(run_id: str | UUID, team_id: int) -> UUID | None:
return TaskRun.objects.filter(id=run_id, team_id=team_id).values_list("task_id", flat=True).first()
+def get_task_run_completed_at(run_ids: Iterable[str | UUID], team_id: int) -> dict[UUID, datetime | None]:
+ """Each run's ``completed_at`` by run id, team-scoped. Runs not found for the team are left out.
+
+ A primary-key read for callers that only need the completion time, so they skip a join to the
+ task run table.
+ """
+ return dict(
+ TaskRun.objects.filter(id__in=list(run_ids), team_id=team_id).order_by().values_list("id", "completed_at")
+ )
+
+
def task_access_scope(*, team_id: int, user: User) -> contracts.TaskAccessScope:
team = Team.objects.get(id=team_id)
access = UserAccessControl(user=user, team=team)