Comment 1 for bug 1804262

Part of the problems seems to be that the image cache manager seems to run on all nodes and checking all instances:

2018-11-20 18:02:13.810 34617 INFO nova.compute.manager [req-b2f865a4-1d07-49d9-aefa-d1e6d0331ff3 - - - - -] Running image cache manager for 705 instances

That's from a debug line I added inside the manager.py and _run_image_cache_manager_pass method:

        LOG.info('Running image cache manager for %d instances', len(filtered_instances))
        self.driver.manage_image_cache(context, filtered_instances)