Parallel pruning

[nijave]

Has parallel pruning ever come up? It's significantly faster for me (almost the same amount of time to prune 100s of snapshots in parallel as it takes to prune a single snapshot)

Usually I do

zfs list -t snapshot -o name my_pool | grep <some criteria> | while read -r snap; do zfs destroy "$snap" & done

Batching might be a middle ground (a 10s or 100s at a time). This would greatly help when reducing retention rules

Looks like it took ~4-5 minutes to prune 1500 snapshots across 2 pools running all in parallel vs ~10 seconds a piece when running sanoid --snapshots-prune

[jimsalterjrs]

Has parallel pruning ever come up? It's significantly faster for me

Interesting! No, hadn't really considered it. I've had to delete thousands of unwanted snapshots from a system before but never tried massive parallelism; I would do your loop with ; done instead of & done.