feat: expose safe source health status

apps/api/app/community_scheduler.py

@@ -16,6 +16,15 @@ from .logging_config import configure_logging
 from .models import CommunityImportRun, DataSource
 
 logger = logging.getLogger("rf4.community_scheduler")
+MAX_BACKOFF_SECONDS = 24 * 60 * 60
+
+def retry_delay(statuses: list[str]) -> int:
+    failures = 0
+    for status in statuses:
+        if status != "failed":
+            break
+        failures += 1
+    return min(settings.community_import_interval_seconds * (2 ** max(0, failures - 1)), MAX_BACKOFF_SECONDS)
 
 def configured_sources():
     return {
@@ -33,8 +42,10 @@ def run_source(source_system: str, *, now: datetime | None = None) -> bool:
         source = session.get(DataSource, source_system)
         if source is None or not source.enabled:
             return False
-        latest = session.scalar(select(CommunityImportRun.started_at).where(CommunityImportRun.source_system == source_system).order_by(CommunityImportRun.started_at.desc()).limit(1))
-        if latest and (latest if latest.tzinfo else latest.replace(tzinfo=timezone.utc)) > current - timedelta(seconds=settings.community_import_interval_seconds):
+        recent = list(session.scalars(select(CommunityImportRun).where(CommunityImportRun.source_system == source_system).order_by(CommunityImportRun.started_at.desc()).limit(8)))
+        latest = recent[0].started_at if recent else None
+        delay = retry_delay([run.status for run in recent])
+        if latest and (latest if latest.tzinfo else latest.replace(tzinfo=timezone.utc)) > current - timedelta(seconds=delay):
             return False
         if session.bind and session.bind.dialect.name == "postgresql" and not session.scalar(text("select pg_try_advisory_xact_lock(hashtext(:key))"), {"key": f"community:{source_system}"}):
             return False