Remove old import task msg on import, and managment command #1368

galaxy/main/management/commands/import_task_message_cleanup.py

@@ -0,0 +1,28 @@
+from django.core.management.base import BaseCommand
+
+from galaxy.main.models import Repository, ImportTaskMessage
+
+
+class Command(BaseCommand):
+    help = ('For each repo, deletes all import_task_messages '
+            'except for messages in latest import_task.')
+
+    def handle(self, *args, **kwargs):
+        total_msg_to_delete = 0
+        total_delete_calls = 0
+
+        repos = Repository.objects.all()
+        for repo in repos:
+            tasks_non_latest = repo.import_tasks.order_by('-id')[1:]
+
+            messages_non_latest = ImportTaskMessage.objects.filter(
+                task__in=tasks_non_latest
+            )
+
+            total_msg_to_delete += messages_non_latest.count()
+            total_delete_calls += 1
+
+            messages_non_latest.delete()
+
+        print('total_msg_to_delete', total_msg_to_delete)
+        print('total_delete_calls', total_delete_calls)

galaxy/worker/tasks.py

@@ -205,6 +205,7 @@ def _import_repository(import_task, logger):
 
     _update_task_msg_content_id(import_task)
     _update_quality_score(import_task)
+    _cleanup_old_task_msg(import_task)
 
     # Updating versions has to go last because:
     # - we don't want to update the version number if the import fails.
@@ -349,6 +350,23 @@ def _update_quality_score(import_task):
     LOG.debug(u'repo quality score: {}'.format(repository.quality_score))
 
 
+def _cleanup_old_task_msg(import_task):
+    old_task_ids = models.ImportTask.objects.filter(
+        repository=import_task.repository,
+    ).exclude(
+        id=import_task.id,
+    ).values_list(
+        'id',
+        flat=True,
+    )
+
+    if old_task_ids:
+        old_task_msgs = models.ImportTaskMessage.objects.filter(
+            task__in=old_task_ids
+        )
+        old_task_msgs.delete()
+
+
 def _update_namespace(repository):
     # Use GitHub repo to update namespace attributes
     if repository.owner.type == 'Organization':