The Linux MD RAID driver when it experiences a bad block on one disk will often incorrectly create bad block entries on a second or third disk in RAID5/RAID6 (https://raid.wiki.kernel.org/index.php/The_Badblocks_controversy).
To check if your member disk has any bad blocks recorded on it run:
# mdadm --examine-badblocks /dev/sdd
To clear the bad blocks list on a fully working drive you can run the following commands:
# mdadm /dev/md127 --fail /dev/sdd --remove /dev/sdd --re-add /dev/sdd --update=force-no-bbl # mdadm /dev/md127 --fail /dev/sdd --remove /dev/sdd --re-add /dev/sdd --update=no-bbl