apache/hadoop · error · IOException

Only found {actualNumINodes} <inode> entries out of {expecte

Error message

Only found {actualNumINodes} <inode> entries out of {expectedNumInodes}

What it means

After the header promises expectedNumINodes, the processor calls expectTag(<inode>) exactly that many times; any IOException from expectTag (typically the unexpected end of the section) is re-thrown with this progress message. It means the XML physically contains fewer <inode> elements than <numInodes> declares - a count/content mismatch, not a parse error.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/tools/offlineImageViewer/OfflineImageReconstructor.java:597

      Long lval = headerNode.removeChildLong(INODE_SECTION_LAST_INODE_ID);
      if (lval != null)  {
        b.setLastInodeId(lval);
      }
      Integer expectedNumINodes =
          headerNode.removeChildInt(INODE_SECTION_NUM_INODES);
      if (expectedNumINodes == null) {
        throw new IOException("Failed to find <numInodes> in INodeSection.");
      }
      b.setNumInodes(expectedNumINodes);
      INodeSection s = b.build();
      s.writeDelimitedTo(out);
      headerNode.verifyNoRemainingKeys("INodeSection");
      int actualNumINodes = 0;
      while (actualNumINodes < expectedNumINodes) {
        try {
          expectTag(INODE_SECTION_INODE, false);
        } catch (IOException e) {
          throw new IOException("Only found " + actualNumINodes +
              " <inode> entries out of " + expectedNumINodes, e);
        }
        actualNumINodes++;
        Node inode = new Node();
        loadNodeChildren(inode, "INode fields");
        INodeSection.INode.Builder inodeBld = processINodeXml(inode);
        inodeBld.build().writeDelimitedTo(out);
      }
      expectTagEnd(INODE_SECTION_NAME);
      recordSectionLength(SectionName.INODE.name());
    }
  }

  private INodeSection.INode.Builder processINodeXml(Node node)
      throws IOException {
    String type = node.removeChildStr(INODE_SECTION_TYPE);
    if (type == null) {
      throw new IOException("INode XML found with no <type> tag.");

View on GitHub (pinned to 2add963021)

Solutions

1. Make counts agree: set <numInodes> to the actual number of <inode> elements (or restore the missing entries)
2. Regenerate the dump and prune with a DOM tool that updates counts automatically
3. Rule out truncation: xmllint --noout plus a look at the tail of the file

Example fix

<!-- before: 5 inodes remain but header still says 8 -->
<INodeSection><numInodes>8</numInodes>...(5 <inode> elements)...</INodeSection>

<!-- after -->
<INodeSection><numInodes>5</numInodes>...(5 <inode> elements)...</INodeSection>

declared=$(xmllint --xpath 'string(//INodeSection/numInodes)' fsimage.xml)
actual=$(xmllint --xpath 'count(//inode)' fsimage.xml)
[ "$declared" = "$actual" ] || { echo "expected $declared inodes, found $actual - fix count or entries"; exit 1; }

When it happens

Trigger: ReverseXML after inode entries were deleted without decrementing <numInodes>; XML truncated mid-section; entries renamed so expectTag(<inode>) no longer matches them (the original exception is chained as the cause).

Common situations: Hand-pruning namespace dumps to remove files; scripted filters that drop entries but not header counts; interrupted copies of the XML.

Related errors

• Failed to find <numInodes> in INodeSection.
• Only read ${actualNumSnapshots} <snapshot> entries out of ${
• Only read ${actualDiffs + 1} diffs out of ${expectedDiffs}
• <createdListSize> was {}, but there were {} <created> entrie
• Only read {} diffs out of {}

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/fc9bb91c2154bdb3. Report an issue: GitHub.