When there are partitions defined on top of a device-mapper device (either a bare device-mapper device or any device-mapper subsystem like LVM or multipath), calling partprobe on such a device causes the original DM UUID to be reused for mappings representing partitions this way: <original_dm_uuid>p<partition_number> This is not very suitable as each subsystem uses its own prefix for DM UUID, for example "LVM-" prefix for LVM devices, "mpath-" for multipath devices, "CRYPT-LUKS1-" for LUKS devices using cryptsetup etc. Using the same DM UUID prefix for mappings representing partitions can cause problems when identifying these devices, for example lsblk then incorrectly identifies the partition mapping as an LVM device if partitions are defined over an LVM device: rawhide/~ # partprobe /dev/vg_test/lvol0 rawhide/~ # dmsetup info -c -o name,uuid Name UUID vg_test-lvol0p1 LVM-uMhMAV4UYSX8yUhYKoiquxOsMQsX24GkCyKytUdEYXK5uexWROeJvH4O4NmpfCjLp1 vg_test-lvol0 LVM-uMhMAV4UYSX8yUhYKoiquxOsMQsX24GkCyKytUdEYXK5uexWROeJvH4O4NmpfCjL rawhide/~ # lsblk -o name,type /dev/vg_test/lvol0 NAME TYPE vg_test-lvol0 (dm-0) lvm `-vg_test-lvol0p1 (dm-1) lvm ...as lsblk derives the exact device type from DM UUID if it's a device-mapper device. This could pose problems in udev rules as well as most of the rules are taking into account DM UUID to resolve the subsystem type used (triggering different parts of the rules). I suggest using the same scheme as kpartx uses nowadays: rawhide/~ # kpartx -a /dev/mapper/vg_test-lvol0 rawhide/~ # lsblk -o name,type /dev/vg_test/lvol0 NAME TYPE vg_test-lvol0 (dm-0) lvm `-vg_test-lvol0p1 (dm-1) part rawhide/~ # dmsetup info -c -o name,uuid Name UUID vg_test-lvol0p1 part1-LVM-uMhMAV4UYSX8yUhYKoiquxOsMQsX24GkCyKytUdEYXK5uexWROeJvH4O4NmpfCjL vg_test-lvol0 LVM-uMhMAV4UYSX8yUhYKoiquxOsMQsX24GkCyKytUdEYXK5uexWROeJvH4O4NmpfCjL rawhide/~ # lsblk -o name,type /dev/vg_test/lvol0 NAME TYPE vg_test-lvol0 (dm-0) lvm `-vg_test-lvol0p1 (dm-1) part Which means using a "partition prefix" instead of suffix in DM UUIDs. Such scheme does not pose any problems for device identification.
Appears in current rawhide (parted-3.1-7.fc19) as well as F18 alpha (parted-3.1-7.fc18). Problem does not appear in F17 (parted-3.0-10.fc17) as UUIDs are not copied and they remain blank for partition mappings. There's still time to include this in F18.
parted-3.1-8.fc18 has been submitted as an update for Fedora 18. https://admin.fedoraproject.org/updates/parted-3.1-8.fc18
Package parted-3.1-8.fc18: * should fix your issue, * was pushed to the Fedora 18 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing parted-3.1-8.fc18' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/FEDORA-2012-16288/parted-3.1-8.fc18 then log in and leave karma (feedback).
parted-3.1-8.fc18 has been pushed to the Fedora 18 stable repository. If problems still persist, please make note of it in this bug report.