What does "Found duplicate PV bUO24Zb2mJGJrTSTRs87Lwrumlu72A1T: using /dev/loop2 not /dev/loop1" mean?
The "Found duplicate PV" message indicates that two volumes (in this case /dev/loop1 and /dev/loop2) contain the same Universally Unique Identification (UUID) label in their LVM metadata. In other words, the volumes /dev/loop1 and /dev/loop2 are identical as far as LVM is concerned. There are several reasons why you would receive this type of messages:
The volumes are in-fact identical! This occurs most often when a SAN is involved (as do practically all "duplicate PV" messages) and is usually because you have multiple paths through your SAN to the same volume. Writes to one volume can be read from the other volume(s) - they appear identical to LVM because they are identical. If this is the case then you should be using device-mapper-multipath or some other multipath I/O (MPIO) solution to enable redundancy. If you are using a multipath solution but still seeing the duplicate PV messages then LVM is not aware of your multipath solution (such as EMC's PowerPath) - in such cases you will need to modify LVM's configuration ("filter = ") so that LVM only uses the multipath device(s) and ignores non-multipath devices. If you fail to do this LVM will effectively bypass your MPIO solution.
One volume is a snapshot or clone of another. See "How do I manage SAN based snapshots of LVM volumes?" for more information.
- Or you've done something wrong, such as:
dd if=/dev/zeroof=/tmp/test.img bs=1M count=10 losetup /dev/loop1 /tmp/test.img pvcreate /dev/loop1 vgcreate test /dev/loop1 lvcreate -l2 -n oops test
Then either something like the commands below, or you've created a snapshot or clone of the volume from the SAN then presented the new snapshot or clone back to this system.
cp /tmp/test.img /tmp/test2.img losetup /dev/loop2 /tmp/test2.img pvscan