apache/hadoop · error · UnsupportedOperationException

Truncate is not supported by BaiduBosFileSystem

Error message

Truncate is not supported by BaiduBosFileSystem

What it means

ZombieJob.convertState() maps the logged Values enum (SUCCESS, FAILED, KILLED) to run-state States and throws IllegalArgumentException('unknown status <v>') for anything else, including null. It is invoked whenever a logged attempt/job result is translated into a simulation state (getTaskAttemptInfo paths). A null Values is by far the most common trigger: rumen JSON traces mark missing data as null, so an attempt record without a 'result' field fails here.

Source

Thrown at hadoop-cloud-storage-project/hadoop-bos/src/main/java/org/apache/hadoop/fs/bos/BaiduBosFileSystem.java:831

            + "file {} before rename", dstPath);
        delete(dstPath, false);
      }
    }

    store.rename(pathToKey(srcPath), pathToKey(dstPath),
        srcStatus.isFile());
    return true;
  }

  @Override
  public Path getWorkingDirectory() {
    return workingDir;
  }

  @Override
  public boolean truncate(Path f, long newLength)
      throws IOException {
    throw new UnsupportedOperationException(
        "Truncate is not supported by BaiduBosFileSystem");
  }

  /**
   * Set the working directory to the given directory.
   *
   * @param newDir the new working directory
   */
  @Override
  public void setWorkingDirectory(Path newDir) {
    workingDir = newDir;
  }

  @Override
  public String getCanonicalServiceName() {
    return null;
  }
}

View on GitHub (pinned to 2add963021)

Solutions

1. Repair or filter the trace: drop attempt records whose 'result' is absent before replay, or re-generate the trace with the rumen tooling so every attempt carries a result.
2. If you control the LoggedJob objects, ensure setResult(...) is always called with SUCCESS/FAILED/KILLED.
3. Guard before simulation: skip attempts with unknown/null result and log them instead of letting the deep convertState() call abort the run.

Example fix

// before
for (LoggedTaskAttempt a : loggedTask.getAttempts()) {
  storyInfo.add(getAttemptInfo(zombieJob, a)); // null result -> IAE in convertState
}

// after
for (LoggedTaskAttempt a : loggedTask.getAttempts()) {
  if (a.getResult() == null
      || !EnumSet.of(Values.SUCCESS, Values.FAILED, Values.KILLED)
             .contains(a.getResult())) {
    LOG.warn("Attempt " + a.getAttemptID() + " has no usable result; skipped");
    continue;
  }
  storyInfo.add(getAttemptInfo(zombieJob, a));
}

static boolean isKnownResult(Values v) {
  return v == Values.SUCCESS || v == Values.FAILED || v == Values.KILLED;
}

static boolean hasUsableResult(LoggedTaskAttempt a) {
  Values r = (a == null) ? null : a.getResult();
  return r == Values.SUCCESS || r == Values.FAILED || r == Values.KILLED;
}

When it happens

Trigger: A LoggedTaskAttempt/LoggedJob whose getResult() is null (missing 'result' in the trace JSON) reaching getTaskAttemptInfo(...)/scaleInfo(...); hand-built LoggedTaskAttempt objects where setResult() was never called; a future/unknown Values constant added by a newer trace format.

Common situations: Replaying rumen traces produced by patched Hadoop versions that omit result for some attempts; traces edited or truncated so records lose fields; unit-test fixtures that build LoggedTaskAttempt without result.

Related errors

• user accessKey and secretAccessKey should not be null
• Can not open a folder
• Failed to rename %s to %s, file already exists or not empty!
• NoSuchUpload
• Key '" + key + "' does not exist in BOS

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/d7da5f65864d6ad3. Report an issue: GitHub.