Use a ConcurrentSkipListMap to track file hashes instead of ConcurrentHashMap. Since the map is a SortedMap, we can easily remove entire subtrees when they are deleted without having to traverse the entire map. Other map operations are now logarithmic time rather than effectively constant time, but I think that's a reasonable tradeoff.
Use a
ConcurrentSkipListMapto track file hashes instead ofConcurrentHashMap. Since the map is aSortedMap, we can easily remove entire subtrees when they are deleted without having to traverse the entire map. Other map operations are now logarithmic time rather than effectively constant time, but I think that's a reasonable tradeoff.