manager: fix "unknown resources cleanup" traceback

When a new on-demand pool is configured in pools.yaml, the cleanup logic
is typically executed before a new resource is started there, and thus
also before the database entry is created.

Fixes: #128
Fixes: #115

resallocserver/manager.py

@@ -688,6 +688,11 @@ def _clean_unknown_resources(self, event):
 
         with session_scope() as session:
             dbinfo = session.query(models.Pool).get(self.name)
+            if not dbinfo:
+                app.log.error("Can't perform cleanup in the configured "
+                              "pool %s: no such pool in database.",
+                              self.name)
+                return
             last_cleanup = dbinfo.cleaning_unknown_resources
             if last_cleanup is None:
                 last_cleanup = datetime.min