I lost my database when my SSD drive went poof last week. My backup is about ~1 TB and I have been backing up my files for about 3 years. My oldest backup is from April 2018. When I reinsalled duplicati on a new drive, I come to find out that the database recreation will take almost 4 months if I leave it running. It is currently processing every one of the 19,292 dblocks. Has been running for the past 12 hours and has only processed about 70 of them. This is after downloading the source files to another local drive ( so there is no netwrok bottleneck). You do the math.
Waiting 4 months to get access to my files again is obviously not practical. I am running on the latest version. All the documentation gives the impression that recreating the DB is a standard procedure and that it should not take long. And then, I come to find out all the different posts on the the same topic of people experiencing the same issue.
Bottom line is that I’m beyond pissed and feel totally duped. Just wanted to share my dissapointment and give everyone else a fair warning. I’m moving on to a different backup solution.
This is correct… it shouldn’t take long, but in some cases it can.
My understanding is it is due to bugs in older versions of Duplicati that caused the dindex files to not be written correctly. Duplicati detects this during a database recreation and it’s forced to download some or all dblocks.
It doesn’t help you at this point, but the most recent Canary versions have solved this issue from what I can tell when you have it rewrite the back end dindex files. (Requires a functioning database to do that.)
When I first started using Duplicati I actually created a secondary backup job whose sole purpose was to back up the database of the first job. If my databases were lost, it would be much quicker to recreate the database of the secondary job, then I could just restore the database of the primary job. But I stopped doing that after using recent canary versions and confirming that database recreations are fast now. (Something I test every few months.)
I don’t know if “currently processing every one of the 19,292 dblocks” was based on any prediction, but
I should also mention (and you might already have seen in some topic) that you can see how major the dblock downloads are likely to be from the progress bar. A ruler in centimeters helps. 90%-100% is bad.
There are three phases, there was a bug fixed in Canary that downloaded dblocks in the earlier phases. You should be able to find detailed information on downloading at About → Show log → Live → Verbose.
For a large backup, it’s possible that you’re still in an early phase (there are three) of dblock downloads, which will end. But at the 90%-100% progress bar, whatever hasn’t been searched is being searched…