fix(api): clean up scan tmp output failure to avoid disk fill (#11421)

Co-authored-by: Pepe Fagoaga <pepe@prowler.com>

api/src/backend/tasks/tasks.py

@@ -467,8 +467,31 @@ def delete_tenant_task(tenant_id: str):
     return delete_tenant(pk=tenant_id)
 
 
+def _scan_tmp_output_directory(tenant_id: str, scan_id: str) -> Path:
+    """Root tmp output directory for a scan ({tmp}/{tenant_id}/{scan_id})."""
+    return Path(DJANGO_TMP_OUTPUT_DIRECTORY) / str(tenant_id) / str(scan_id)
+
+
+class ScanReportRLSTask(RLSTask):
+    """
+    RLS task that removes the scan's tmp output directory when the task fails.
+
+    Covers failures both inside and outside the task body (e.g. ENOSPC mid-write,
+    or setup errors) so partial artifacts do not accumulate on the worker disk.
+    """
+
+    def on_failure(self, exc, task_id, args, kwargs, _einfo):  # noqa: ARG002
+        del args  # Required by Celery's Task.on_failure signature; not used.
+        tenant_id = kwargs.get("tenant_id")
+        scan_id = kwargs.get("scan_id")
+
+        if tenant_id and scan_id:
+            logger.error(f"Scan report task {task_id} failed: {exc}")
+            rmtree(_scan_tmp_output_directory(tenant_id, scan_id), ignore_errors=True)
+
+
 @shared_task(
-    base=RLSTask,
+    base=ScanReportRLSTask,
     name="scan-report",
     queue="scan-reports",
 )
@@ -518,6 +541,9 @@ def generate_outputs_task(scan_id: str, provider_id: str, tenant_id: str):
     out_dir, comp_dir = _generate_output_directory(
         DJANGO_TMP_OUTPUT_DIRECTORY, provider_uid, tenant_id, scan_id
     )
+    # Removed on success here and on failure by ScanReportRLSTask.on_failure,
+    # so partial artifacts do not accumulate and fill the disk (ENOSPC).
+    scan_tmp_dir = _scan_tmp_output_directory(tenant_id, scan_id)
 
     def get_writer(writer_map, name, factory, is_last):
         """
@@ -666,7 +692,7 @@ def generate_outputs_task(scan_id: str, provider_id: str, tenant_id: str):
         # TODO: We need to create a new periodic task to delete the output files
         # This task shouldn't be responsible for deleting the output files
         try:
-            rmtree(Path(compressed).parent, ignore_errors=True)
+            rmtree(scan_tmp_dir, ignore_errors=True)
         except Exception as e:
             logger.error(f"Error deleting output files: {e}")
         final_location, did_upload = upload_uri, True

api/src/backend/tasks/tests/test_tasks.py

@@ -15,8 +15,10 @@ from tasks.jobs.lighthouse_providers import (
 from tasks.tasks import (
     DJANGO_TMP_OUTPUT_DIRECTORY,
     STALE_TMP_OUTPUT_MAX_AGE_HOURS,
+    ScanReportRLSTask,
     _cleanup_orphan_scheduled_scans,
     _perform_scan_complete_tasks,
+    _scan_tmp_output_directory,
     check_integrations_task,
     check_lighthouse_provider_connection_task,
     generate_outputs_task,
@@ -771,6 +773,38 @@ class TestGenerateOutputs:
             mock_s3_task.assert_called_once()
 
 
+class TestScanReportRLSTaskOnFailure:
+    def test_on_failure_removes_scan_tmp_directory(self):
+        task = ScanReportRLSTask()
+
+        with patch("tasks.tasks.rmtree") as mock_rmtree:
+            task.on_failure(
+                exc=OSError("No space left on device"),
+                task_id="task-abc",
+                args=(),
+                kwargs={"tenant_id": "t-1", "scan_id": "s-1"},
+                _einfo=None,
+            )
+
+        mock_rmtree.assert_called_once_with(
+            _scan_tmp_output_directory("t-1", "s-1"), ignore_errors=True
+        )
+
+    def test_on_failure_skips_when_missing_kwargs(self):
+        task = ScanReportRLSTask()
+
+        with patch("tasks.tasks.rmtree") as mock_rmtree:
+            task.on_failure(
+                exc=OSError("No space left on device"),
+                task_id="task-abc",
+                args=(),
+                kwargs={},
+                _einfo=None,
+            )
+
+        mock_rmtree.assert_not_called()
+
+
 class TestScanCompleteTasks:
     @patch("tasks.tasks.aggregate_attack_surface_task.apply_async")
     @patch("tasks.tasks.chain")