I couldn’t find my original post so I tested the database Repair process again (using 2.0.3.9 canary) on a very small (200M source) backup with the following results:
- Deleting all 17
.dindexfiles and running REPAIR resulted in 17 fresh (dated today).dindexfiles being created in just over 1 minute - Deleting all 18
.dlistfiles and running REPAIR resulted in 18 fresh (dated today).dlistfiles being created in just over 1 minute (though ONE file was flagged with a warning of being uploaded with size X but should be size Y) - Deleting all 17
.dindexAND all 18.dlistfiles and running REPAIR resulted in 35 fresh (dated today) files being created in 1.5 minutes
So I’d say yes - at least with 2.0.3.9 it is safe to delete all .dindex and/or .dlist files as they can be recreated from the local database ASSUMING the local database is in good condition.
My GUESS is that even if a user has no local database it would still be safe to do the tests I just did, they would simply take longer as all the .dblock files would have to be downloaded and parsed to recreate the missing .dlist and/or .dindex files.
Let me know if you’d like me to test that as well.