the temporary snaps we create when doing fastdiff can accidentally end up being sent around, and when two of them get caught, there's a name collision because you can't have two snaps with the same name in zfs - fix this by deleting the snaps before we send the filesystem anywhere, and in the intra-cluster case where we can't do that because receiving is async there, at least cleaning them up after we receive a filesystem.
the temporary snaps we create when doing fastdiff can accidentally end up being sent around, and when two of them get caught, there's a name collision because you can't have two snaps with the same name in zfs - fix this by deleting the snaps before we send the filesystem anywhere, and in the intra-cluster case where we can't do that because receiving is async there, at least cleaning them up after we receive a filesystem.