Insufficient suitable allocatable extents error when performing RAID takeover

To improve performance while still keeping redundancy, I am trying to convert the root file system of a NAS from a RAID 5 logical volume to a RAID 10 logical volume, while keeping the same number of disks used.

In this section of the lvmraid man page, it suggests converting RAID 10 and RAID 5 is possible with an indirect conversion.

The following takeover conversions are currently possible:

...

• between striped/raid0 and raid5.

...

• between striped/raid0 and raid10.

But when running the appropriate lvconvert command (on a similar test LV), repeatedly as prompted to perform the intermediate conversions, I get this error in the last step when finally converting from RAID 0 to RAID 10:

root@nas:~# lvcreate --type raid5 -i 3 -n test -L 100M vg
  Using default stripesize 64.00 KiB.
  Rounding size 100.00 MiB (25 extents) up to stripe boundary size 108.00 MiB (27 extents).
  Logical volume "test" created.

root@nas:~# lvconvert vg/test --type raid10 -m 1 -i 2
  Using default stripesize 64.00 KiB.
  Replaced LV type raid10 (same as raid10_near) with possible type raid5_n.
  Repeat this command to convert to raid10 after an interim conversion has finished.
  Converting raid5 (same as raid5_ls) LV vg/test to raid5_n.
Are you sure you want to convert raid5 LV vg/test? [y/n]: y
  Logical volume vg/test successfully converted.

root@nas:~# lvconvert vg/test --type raid10 -m 1 -i 2
  Using default stripesize 64.00 KiB.
  Replaced LV type raid10 (same as raid10_near) with possible type raid0_meta.
  Repeat this command to convert to raid10 after an interim conversion has finished.
Are you sure you want to convert raid5_n LV vg/test to raid0_meta type? [y/n]: y
  Logical volume vg/test successfully converted.

root@nas:~# lvconvert vg/test --type raid10 -m 1 -i 2
  Using default stripesize 64.00 KiB.
Are you sure you want to convert raid0_meta LV vg/test to raid10 type? [y/n]: y
  Insufficient suitable allocatable extents for logical volume : 30 more required

root@nas:~#

Even when freeing up physical extents and reducing the LV size, I get the same error. All the physical volumes are allocatable.

Here is the vgdisplay output:

  --- Volume group ---
  VG Name               vg
  System ID             
  Format                lvm2
  Metadata Areas        4
  Metadata Sequence No  333
  VG Access             read/write
  VG Status             resizable
  MAX LV                0
  Cur LV                3
  Open LV               2
  Max PV                0
  Cur PV                4
  Act PV                4
  VG Size               <3.64 TiB
  PE Size               4.00 MiB
  Total PE              953864
  Alloc PE / Size       943654 / <3.60 TiB
  Free  PE / Size       10210 / 39.88 GiB
  VG UUID               PF08Ce-xvsS-7KsA-0Vw6-zkXR-jyUf-ocDWDW

How do I resolve this error? I would like to use the same PEs for the converted LV, shrinking the file system as necessary.

For the purposes of this description, consider disk to mean a disk, partition or other logical container. Each disk is considered to be the same size, holding a unit of storage (1GB, 1TB, whatever; it doesn't matter).

• RAID 5 uses three disks for two units of useful storage
• RAID 0 uses two disks for two units of useful storage
• RAID 10 uses two disks for one unit of useful storage; typically RAID 10 can be considered as RAID 1 on top of RAID 0 (or v.v.) so you would use four disks for two units of useful storage

Also

• When you convert from RAID 5 to RAID 0 you have one unit of storage left over (i.e. one disk is spare)
• When you convert from RAID 0 to RAID 10 you need an additional unit of storage for each unit that you're converting (i.e. two extra disks on top of the two you're currently using)

Consequence: you started with three disks so you're one disk short.

One option might be to see if you can reduce your RAID 0 array down from using two disks to one, but this presupposes your original RAID 5 array is only half full. The other option is to obtain a fourth disk.