zdb block size statistics for individual datasets

[jintakhan]

Describe the feature would like to see added to OpenZFS

zdb -Lbbbs zpool/dataset should calculate a block size histogram for datasets like it does when run on a zpool instead.

How will this feature improve OpenZFS?

Currently zdb -Lbbbs only returns a single line with very little information and statistics when run on a zpool: Dataset zpool/dataset [ZPL], ID 12345, cr_txg 1234567, 123.45G, 12345 objects

The command would be much more helpful in tuning datasets on a zpool, for example, to different record sizes and special device allocation sizes if the data breakdown was able to return the full block size statistics of a dataset.

Additional context

Others have been asking about this in other forums:

• https://discourse.practicalzfs.com/t/zdb-lbbbs-poolname-equivalent-for-dataset/396
• Similar, but with dedup in mind: https://www.reddit.com/r/zfs/comments/rtvr35/zdb_on_dataset_possible/