I’ve got a large (multi-TB) backup I’m attempting via Duplicati to Jottacloud using the latest Beta (2.0.3.3). This will take many weeks at my current upload speed (here’s hoping the multi-threaded features get implemented in Beta soon and can speed this up). If my computer crashes during the backup, or I need to reboot for one reason or another, can I still restore from a backup that I canceled while it was in progress, or does the full backup need to be completed to be able to restore any files?
That’s a good question - I’ve never tested a partial initial backup before.
My GUESS is that if the cancel is done politely (meaning use the cancel button and choose “after next upload”) things will work as expected, but don’t take my word on that.
I’ll try to simulate that scenario in the next few days and get back to you.
The .dlist file is not produced until the end of the backup run. This file is the “restore point” that Duplicati references to figure out what files can be restored. It’s, in a sense, just a list of all the file names and what blocks the file can be constructed from.
So on a new unfinished backup Duplicati cannot restore any data. See next comment 
This also goes for newer backup runs. If your machine crashes before finishing it’s 2nd backup run then you can only restore files from the 1st run.
I tried to replicate this, but I think my Duplicati server is stubborn… When I choose “after next upload” it just keeps going until it finished uploading all the data…
I definitely remember this option once stopping after the next volume… (Edit: it’s been “stopping” for 20 minutes and has produced almost 100 dblock and dindex files 
)
In any case I would expect “after next upload” to also not work, because it would stop after the volume without adding the .dlist file (otherwise you’d have a partial snapshot every time you cancel your backup)
Haha, I’ve been proven wrong again after ending my test.
It turns out that the partial dlist is sitting in your local DB, so provided that you can still use your system, or extract this database, you can infact restore data 
I have no .dlist on the remote, so it’s important to note that you’re really depending on the local DB then.
1 Like
Thanks for doing that test! It’s probably more effort than I’d like, but it would be nice if polite cancel placed the partial dlist remotely as well, otherwise people with large sources or show destinations aren’t “fully” covered until the first backup is done.
Though I suppose in the case of a system failure before initial backup is finished a database restore would use the dblocks that could be found and likely allow a restore from that…