apache/hadoop · error · NullPointerException

key == null

Error message

key == null

What it means

LightWeightGSet, Hadoop's chained hash set behind hot structures like the NameNode INode map, has no null-key semantics: get(K) throws NullPointerException("key == null") before any hashing. The explicit check fails fast with a searchable message instead of an opaque NPE from inside the hashing logic.

Source

Thrown at hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/LightWeightGSet.java:130

  public int size() {
    return size;
  }

  protected int getIndex(final K key) {
    return key.hashCode() & hash_mask;
  }

  protected E convert(final LinkedElement e){
    @SuppressWarnings("unchecked")
    final E r = (E)e;
    return r;
  }

  @Override
  public E get(final K key) {
    //validate key
    if (key == null) {
      throw new NullPointerException("key == null");
    }

    //find element
    final int index = getIndex(key);
    for(LinkedElement e = entries[index]; e != null; e = e.getNext()) {
      if (e.equals(key)) {
        return convert(e);
      }
    }
    //element not found
    return null;
  }

  @Override
  public boolean contains(final K key) {
    return get(key) != null;
  }

View on GitHub (pinned to 2add963021)

Solutions

1. Use the stack trace to find the caller passing null and null-check the identifier at its source.
2. Give null a real meaning at the boundary: skip the lookup or throw a descriptive domain exception.
3. Never store or query null keys in GSet-based structures.

Example fix

// before
INode node = inodes.get(pathComponent); // pathComponent may be null

// after
INode node = (pathComponent != null) ? inodes.get(pathComponent) : null;

if (key == null) {
  return null; // or throw a domain exception describing the missing identifier
}
return gset.get(key);

When it happens

Trigger: Calling gset.get(null) - a null identifier (inode name, block id) propagated from a failed parse or an upstream lookup that returned null.

Common situations: Refactors that made an identifier optional; a lookup-by-name where the name was never validated; code assuming the set tolerates null keys like java.util.HashMap.

Related errors

• Null element is not supported.
• !(element instanceof LinkedElement), element.getClass()=" +
• modification=" + modification + " != iterModification = " +
• There are no more elements
• There is no current element to remove

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/4e622ed28537f630. Report an issue: GitHub.