Long story short, it was proven to be not a bug but rather an incorrect test.
In some cases, we write to the sector already being tracked, which results in a
non-incrementing nr_changed_blocks counter. That happened because a small
amount of data was written (about 15 bytes) and the chances of writing to the
tracked sector were relatively high. The fix is to write more to increase the probability
of writing to the sector which isn't tracked.
Long story short, it was proven to be not a bug but rather an incorrect test.
In some cases, we write to the sector already being tracked, which results in a non-incrementing
nr_changed_blockscounter. That happened because a small amount of data was written (about 15 bytes) and the chances of writing to the tracked sector were relatively high. The fix is to write more to increase the probability of writing to the sector which isn't tracked.Closes #167