apache/druid · error · IllegalArgumentException

Cannot add files of the same name, already have

Error message

Cannot add files of the same name, already have [%s]

What it means

Each file inside a SegmentFileV10 container must have a unique name because the container metadata indexes internal files by name. addWithChannel checks internalFiles.containsKey(name) and throws this IllegalArgumentException when a duplicate name is added. It prevents silently overwriting or shadowing an existing internal file.

Solutions

1. Make internal file names unique by adding a distinguishing suffix (column name, version, index kind) before calling addWithChannel
2. Ensure a single SegmentFileBuilderV10 instance is not reused to write the same file twice; create a new builder or a new container for a second write
3. Check existing names with the builder's metadata before adding and skip/overwrite intentionally

Example fix

// before
builder.addWithChannel("__time.idx", size);
builder.addWithChannel("__time.idx", size2); // duplicate
// after
builder.addWithChannel("__time.idx.v1", size);
builder.addWithChannel("__time.idx.v2", size2);

// Java: track names you have already added
Set<String> used = new HashSet<>();
if (!used.add(name)) { throw new IllegalStateException("duplicate file name: " + name); }

try {
  builder.addWithChannel(name, size);
} catch (IllegalArgumentException e) {
  if (e.getMessage().startsWith("Cannot add files of the same name")) {
    builder.addWithChannel(appendVersionSuffix(name), size);
  } else { throw e; }
}

When it happens

Trigger: Calling addWithChannel(name, size) (directly or via add/makeColumn) twice with the same name within one container build, e.g. writing two versions of a column under the same file name in one builder session.

Common situations: Custom ingestion/segment-generation code writing multiple index files with the same generated name; a builder reused across bundles without clearing names; a bug in name generation omitting a version/suffix.

Understand the failure class

Background: "already exists" / EEXIST / FileAlreadyExistsException: what the 'file already exists' error means and how to fix it — this error's family across 37 libraries.

Related errors

• Cannot have a comma in the name of a file, got
• A local input source accepts only one of
• A local input source can set parameter
• A local input source requires one parameter of
• A local input source requires one property of

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/4581cd40c3fd5084. Report an issue: GitHub.

Appendix: source

Thrown at processing/src/main/java/org/apache/druid/segment/file/SegmentFileBuilderV10.java:203

    }
  }

  @Override
  public void add(String name, ByteBuffer bufferToAdd) throws IOException
  {
    try (SegmentFileChannel out = addWithChannel(name, bufferToAdd.remaining())) {
      out.write(bufferToAdd);
    }
  }

  @Override
  public SegmentFileChannel addWithChannel(final String name, final long size) throws IOException
  {
    if (name.contains(",")) {
      throw new IAE("Cannot have a comma in the name of a file, got[%s].", name);
    }
    if (internalFiles.containsKey(name)) {
      throw new IAE("Cannot add files of the same name, already have [%s]", name);
    }
    ensureNameMatchesActiveBundle(name);
    if (size > maxContainerSize) {
      throw DruidException.forPersona(DruidException.Persona.ADMIN)
                          .ofCategory(DruidException.Category.RUNTIME_FAILURE)
                          .build(
                              "Serialized buffer size[%,d] for column[%s] exceeds the maximum[%,d]. "
                              + "Consider adjusting the tuningConfig - for example, reduce maxRowsPerSegment, "
                              + "or partition your data further.",
                              size, name, maxContainerSize
                          );
    }

    // If an outer writer is mid-write we can't append to the current container concurrently, route through a temp
    // file that will be merged back into a container once the outer writer releases.
    if (writerCurrentlyInUse) {
      return delegateChannel(name, size);
    }

View on GitHub (pinned to 9b90983fd2)