apache/hadoop · error · NameNodeFormatException

NameNode format aborted as reformat is disabled for this clu

Error message

NameNode format aborted as reformat is disabled for this cluster.

What it means

During 'hdfs namenode -format', if dfs.reformat.disabled=true and any storage directory already has data (sd.hasSomeData()), NameNode.format throws NameNodeFormatException. It is a deliberate safety guard: -force and interactive prompts are neutralized (force=false, isInteractive=false) so an existing cluster's metadata can never be wiped while the guard is on.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/NameNode.java:1413

      //Generate a new cluster id
      clusterId = NNStorage.newClusterID();
    }

    LOG.info("Formatting using clusterid: {}", clusterId);
    FSNamesystem fsn = null;
    try (FSImage fsImage = new FSImage(conf, nameDirsToFormat, editDirsToFormat)) {
      fsn = new FSNamesystem(conf, fsImage);
      fsImage.getEditLog().initJournalsForWrite();

      // Abort NameNode format if reformat is disabled and if
      // meta-dir already exists
      if (conf.getBoolean(DFSConfigKeys.DFS_REFORMAT_DISABLED,
          DFSConfigKeys.DFS_REFORMAT_DISABLED_DEFAULT)) {
        force = false;
        isInteractive = false;
        for (StorageDirectory sd : fsImage.storage.dirIterable(null)) {
          if (sd.hasSomeData()) {
            throw new NameNodeFormatException(
                "NameNode format aborted as reformat is disabled for "
                    + "this cluster.");
          }
        }
      }

      if (!fsImage.confirmFormat(force, isInteractive)) {
        return true; // aborted
      }

      fsImage.format(fsn, clusterId, force);
    } catch (IOException ioe) {
      LOG.warn("Encountered exception during format", ioe);
      throw ioe;
    } finally {
      if (fsn != null) {
        fsn.close();
      }

View on GitHub (pinned to 2add963021)

Solutions

1. If reformat is genuinely intended, temporarily set dfs.reformat.disabled=false in hdfs-site.xml, run the format, then restore it to true.
2. Otherwise decommission properly: verify the metadata is disposable, then move/clear the contents of dfs.namenode.name.dir before formatting (or format into fresh, empty directories).
3. For HA clusters, format on one node and bootstrap the rest with 'hdfs namenode -bootstrapStandby' instead of formatting everywhere.

Example fix

<!-- before: hdfs-site.xml -->
<property><name>dfs.reformat.disabled</name><value>true</value></property>
hdfs namenode -format   # aborts: NameNode format aborted as reformat is disabled

<!-- after -->
<property><name>dfs.reformat.disabled</name><value>false</value></property>
hdfs namenode -format   # succeeds; then set the guard back to true

// guard a scripted format against the reformat lock
boolean reformatDisabled = conf.getBoolean("dfs.reformat.disabled", false);
boolean hasData = Arrays.stream(nameDir.listFiles())
    .anyMatch(f -> f.getName().matches("current|in_use.lock|previous|seen_txid"));
if (reformatDisabled && hasData)
  throw new IllegalStateException("refusing to format: dfs.reformat.disabled=true and data exists");

boolean isReformatGuard(Throwable t) {
  return t instanceof NameNodeFormatException
      && String.valueOf(t.getMessage()).contains("reformat is disabled");
}

When it happens

Trigger: Running 'hdfs namenode -format' (with or without -force) on a node where dfs.reformat.disabled=true in hdfs-site.xml and any name dir contains existing fs image/edits/in_use.lock data.

Common situations: Production hardening sets dfs.reformat.disabled=true; months later an operator or an automation playbook re-runs the format step and hits the guard; re-provisioning scripts that format on every run.

Related errors

• The option dfs.namenode.support.allow.format is set to false
• Unknown nameservice: {}
• Configuration has multiple addresses that match local node's
• Configuration dfs.namenode.rpc-address must be suffixed with
• Unexpected configuration parameters: dfs.namenode.replicatio

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/e7e8ce7a3844e404. Report an issue: GitHub.