Hi, this has been fixed upstream in v2.03.17.
As a workaround you can create the shared VG with an extent size that is a) less or equals to 256M and b) a divisible of your target extent size. Then extent the lvmlock LV so it is also a whole number of extents with your target extent size. And finally change the extent size of the VG. Obviously all the caveats of changing the extent size of a VG still apply. root@node1:~# vgcreate --shared --physicalextentsize 256M my-shared-test-vg /dev/sdb1 Enabling sanlock global lock Physical volume "/dev/sdb1" successfully created. Logical volume "lvmlock" created. Volume group "my-shared-test-vg" successfully created VG my-shared-test-vg starting sanlock lockspace Starting locking. Waiting until locks are ready... root@node1:~# vgs -o vg_name,vg_shared,vg_size,vg_extent_size,vg_extent_count,vg_free,vg_free_count VG Shared VSize Ext #Ext VFree Free my-shared-test-vg shared 4.00g 256.00m 16 3.75g 15 root@node1:~# lvextend --size 1G /dev/mapper/my--shared--test--vg-lvmlock Size of logical volume my-shared-test-vg/lvmlock changed from 256.00 MiB (1 extents) to 1.00 GiB (4 extents). Logical volume my-shared-test-vg/lvmlock successfully resized. root@node1:~# vgchange --physicalextentsize 1G my-shared-test-vg Volume group "my-shared-test-vg" successfully changed root@node1:~# vgs -o vg_name,vg_shared,vg_size,vg_extent_size,vg_extent_count,vg_free,vg_free_count VG Shared VSize Ext #Ext VFree Free my-shared-test-vg shared 4.00g 1.00g 4 3.00g 3 root@node1:~# lvs -a LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert [lvmlock] my-shared-test-vg -wi-ao---- 1.00g This can still be a bit finicky because you have to get the size of the PV right, so the initial extent count is a multiple of the final extent count. In the above example /dev/sdb1 is 4200MiB, which is 16 and 4 extents respectively. It does not work with 4300MiB, because that would be 17 and 4; vgchange will show the error "New extent size is not a perfect fit". --- Greetings Corubba
