Description
The issue is that bucket count is used for checking the scaled down size of the hash tables but is used later on to convert to the map join as well which may be incorrect in cases where the entire hash table does not fit in the specified size.