I just spent the whole morning debugging a problem that I thought was in my backup script. Seems that the drive I use to hold backups of /home was full, even though I had just cleaned it out a couple of weeks ago.
Every daily backup was taking something like 20 GB of NEW hard drive space - and I can assure you that I don't make that many changes in my home directory over the course of a day. At 20 GB per day, it wasn't taking long to fill my 230 GB partition, especially since there are 2 virtual machines that live on that partition as well.
So, I hunted for the problem. And I hunted. And I hunted.
And I found it.
Nepomuk.
Seems that my semantic search database has achieved the status of nearly 20 GB of space on my /home partition. This for a system that doesn't even work reliably.
The file ~/.kde4/share/apps/nepomuk/repository/main/data/virtuosobackend/soprano-virtuoso.db is a paltry 19415433216 bytes in size. And, of course, it does change every day...sometimes substantially on those days where virtuoso actually decides to run reliably. So the whole thing gets copied every night.
So I excluded the directory ~/.kde4/share/apps/nepomuk from the backup, and went through the backup directory and deleted that directory from every backup, and now I have 163 GB available on that partition.
A 20GB nepomuk file. There's something obscene about that. But it is something to know about; your backups will go a lot better if you exclude that file.