I am testing a small ceph cluster implemented in qemu-kvm and seeing there are a few cloudy types here who may have tried this, I though I'd ask about my problem before reworking.
I decided to go btrfs in the VM's and the ceph storage (all qcow2) on a reiserfs host fs. So far so good, but I am getting btrfs problems - throws an error and goes readonly -> oops/crash in btrfs I am now using cache=none for qemu (which is recommended for vm's on a btrfs host, though not the case here) and that helped some. My question is ... are there any other settings to help this configurations stability/reliability using in vm btrfs? Otherwise I'll go back to reiserfs for the hosts and try xfs for the ceph storage. BillK
