One of my quick-and-dirty gotos for getting a rough idea of buffered-writes size + disk-write activity on random linux systems is: `watch -n1 grep -ie dirty -e writeback /proc/meminfo`.

You can invoke `sync` to watch the buffered-writes queue burn down when you have lots of pending writes.

see: `LESS=+/meminfo man proc` or https://github.com/torvalds/linux/blob/master/Documentation/... for more info