We have installed VDR 2.0 about 5 months ago and have about 20 small VMs being backed up.
According to the daily status report mail, the used storage is about 340 GB. This value is stable.
However, when checking the disk usage on the CIFS server (Windows 2003 server), it is about
690 GB, and this is ever increasing.
I think this reflects the partial usage of the 1GB sized files in the destination directory.
When I installed this VDR 2.0 I started with a clean deduplication store. However, I have also
used earlier versions. With earlier versions I noticed that slab-### files were created with
sequential numbers as space requirement increased, but with the 2.0 version the files are
created with random numbers. It suggests that usage of the store is now done via some hashing,
instead of sequentially from the start.
It appears that the daily Reclaim task, which successfully deletes older backups from the store,
frees up blocks in the deduplication store, but those blocks are not used by new backups.
Instead, new slab-### files are being created all the time.
The older VDR versions did not have this problem. Although the size of the store always had
the tendency to increase, it did not grow as rapidly as the 2.0 version does.
Do others observe this behaviour? Is there anything that can be done about it?
We don't want to use all the space on the CIFS server, but rather keep it limited to what the
VDR actually requires to store the backups.
(we keep a backup of the deduplication store for disaster recovery, and this takes more and
more time to complete)
