Update: changing the fleecing setting didn't change anything, the backup crashed 50% into 3TB and the VM was unusable, I had to stop/start, which did not work at all.As all VMs concerned had the ballooning flag set, most of them to 50%. I removed...
Interestingly, everything works fine for me on 7 pbs/pve combinations recently updated to latest. Only 1 (!) SLES12 instance experiences this problem every single day during backup.
If I start the backup job manually it seems to work fine...