Description of problem: This is likely the same issue as bug 481384. [root@hayes-03 cache]# lvs -a -o +devices LV VG Attr LSize Origin Snap% Move Log Copy% Convert Devices origin1 base1 owi-a- 100.00M /dev/etherd/e1.1p1(0) snap1 base1 swi-a- 12.00M origin1 0.13 /dev/etherd/e1.1p1(25) origin2 base2 owi-a- 100.00M /dev/etherd/e1.1p2(0) snap2 base2 swi-a- 12.00M origin2 0.13 /dev/etherd/e1.1p2(25) [root@hayes-03 cache]# pvcreate /dev/base1/origin1 Physical volume "/dev/base1/origin1" successfully created [root@hayes-03 cache]# pvcreate /dev/base2/origin2 Physical volume "/dev/base2/origin2" successfully created [root@hayes-03 cache]# pvscan Found duplicate PV F4qW107N5AhBuGaVkFRgLFYCbjhEf7Ms: using /dev/mapper/base1-origin1-real not /dev/base1/origin1 Found duplicate PV rh9vKtdPUAcItGPnmI7OyNdopfI2WSpi: using /dev/mapper/base2-origin2-real not /dev/base2/origin2 PV /dev/etherd/e1.1p2 VG base2 lvm2 [1.77 TB / 1.77 TB free] PV /dev/etherd/e1.1p1 VG base1 lvm2 [1.77 TB / 1.77 TB free] PV /dev/sda2 VG VolGroup00 lvm2 [74.38 GB / 0 free] PV /dev/mapper/base1-origin1-real lvm2 [100.00 MB] PV /dev/mapper/base2-origin2-real lvm2 [100.00 MB] PV /dev/etherd/e1.1p3 lvm2 [1.77 TB] PV /dev/etherd/e1.1p4 lvm2 [1.77 TB] PV /dev/etherd/e1.1p5 lvm2 [1.77 TB] Total: 8 [8.94 TB] / in use: 3 [3.62 TB] / in no VG: 5 [5.32 TB] Version-Release number of selected component (if applicable): 2.6.18-128.el5 lvm2-2.02.40-6.el5 BUILT: Fri Oct 24 07:37:33 CDT 2008 lvm2-cluster-2.02.40-7.el5 BUILT: Wed Nov 26 07:19:19 CST 2008 device-mapper-1.02.28-2.el5 BUILT: Fri Sep 19 02:50:32 CDT 2008 How reproducible: Everytime
Removing 5.4 flag. Stacking comes in two flavors: 1) The ability to stack LVs in the same vg, like creating a RAID10, etc 2) Creating new VGs on top of existing LVs in another VG. #1 is a significant portion of work that has been talked about for a long time, but probably won't be seen in RHEL5. #2 is possible in rudimentary forms. When more exotic stacking takes place - like putting a VG on top of a mirror or snapshot - the concept starts to break down. It is possible to run around and stomp out all the bugs associated with this in the RHEL5 timeframe, but I don't think anyone can be tasked to it in 5.4.
Filter the snapshots in lvm.conf and it will work. Not a bug, but an incorrect configuration.
If stacking on top of an origin+snapshot, the snapshot *must* be writeable and the PV uuid within it changed OR filters must be used to ignore one of the two devices. It is impossible for any general rule to guess the right default without some level of manual intervention.
Maybe we could change the error message to suggest to the user that he should edit "filter" clause in lvm.conf --- because this problem is repeated several times in our bugzilla.