LinuxQuestions.org - OpenZFS on Linux and BSD zdb command gives space map refcount mismatch

I recently found the zdb command and after running it showed these errors.

Code:

user@Hostname:~# zdb -d zfs_pool

Dataset mos [META], ID 0, cr_txg 4, 16.8M, 189 objects

Dataset zfs_pool/test2 [ZPL], ID 106, cr_txg 166647, 30.0G, 7 objects

Dataset zfs_pool/test [ZPL], ID 131, cr_txg 166799, 30.0G, 7 objects

Dataset zfs_pool/test4 [ZPL], ID 157, cr_txg 175795, 1.45M, 7 objects

Dataset zfs_pool/test3 [ZPL], ID 151, cr_txg 175783, 2.51M, 7 objects

Dataset zfs_pool [ZPL], ID 21, cr_txg 1, 96.0K, 8 objects



Verified large_blocks feature refcount is correct (2)

space map refcount mismatch: expected 98 != actual 58

Note: Datasets have been removed from code text to keep it simple.

This was run on Ubuntu Server 17.10 (4.13.0-21-generic #24-Ubuntu) and ZFS 0.6.5.11-1
Also on a zfs pool that has barely been used and was created about 4 months ago with brand new
WD Re4 drives(312 Hours so far). And never been any errors on zpool scrub.

Is this "space map refcount mismatch" bad ?

After finding this out I ran it on a Completely different system running FreeNAS-11.1-RELEASE

Code:

[user@Hostname2 ~]# zdb -d -U /data/zfs/zpool.cache -C NAS_Pool

Dataset NAS_Pool/jails [ZPL], ID 188, cr_txg 5408, 128K, 7 objects

Dataset NAS_Pool/Backups [ZPL], ID 103, cr_txg 2988, 128K, 11 objects

Dataset NAS_Pool [ZPL], ID 21, cr_txg 1, 140K, 12 objects

Verified large_blocks feature refcount of 0 is correct

Verified sha512 feature refcount of 0 is correct

Verified skein feature refcount of 0 is correct

space map refcount mismatch: expected 66 != actual 42

Note: Datasets and some text have been removed from code text to keep it simple

This was run a fairly new pool with very little use and never any scrub errors too.

Searching the net it seems like this maybe a problem in ZFS or the zdb program itself.
Anyone know what this is ? If its bad how can it be prevented ?

Does your pool give you any errors ?