There is always a non-zero possibility that a further incident occurs while resilvering after a first incident. (And there are many possibilities for “incident”…)
Managing disks of different capacities is possible, but mixing different vdev geometries within a pool is not recommended.
From your nine drives, you could have 3*2 TB, 3*4 TB and 8+3+3 TB in three raidz1 vdevs (I would strongly prefer raidz2 but that’s another story…), and it would be fine—save for losing 5 TB on the single 8 TB drive until both 3 TB drives are replaced by larger drives.