Details
Description
Our logic of later deletion of db files doesn't work if node is ejected and forgets about non-standard db path it had. If that was the case it cannot know where old DB files are thus it's unable to delete them freeing space.
NOTE: 2.0 as of now actually preserves it's db path after failover. But not 1.8.1 and I'd consider this preservation a 2.0 bug. Even if not, people can specify different DB path at node re-initialization and that'll also lead to losing track of old db files.