apache/flink · error · RuntimeException

Unsupported page type: {}

Error message

Unsupported page type: {}

What it means

RuntimeException from the read loop in AbstractColumnReader.readToVector: after finishing a page, pageReader.readPage() returned an object that is neither DataPageV1 nor DataPageV2. parquet-mr's page model has only these two concrete data page formats, so this branch indicates an unexpected page implementation - practically only seen with version-skewed parquet-mr jars on the classpath or corrupt page headers.

Source

Thrown at flink-formats/flink-parquet/src/main/java/org/apache/flink/formats/parquet/vector/reader/AbstractColumnReader.java:160

    /** Reads `total` values from this columnReader into column. */
    @Override
    public final void readToVector(int readNumber, VECTOR vector) throws IOException {
        int rowId = 0;
        WritableIntVector dictionaryIds = null;
        if (dictionary != null) {
            dictionaryIds = vector.reserveDictionaryIds(readNumber);
        }
        while (readNumber > 0) {
            // Compute the number of values we want to read in this page.
            int leftInPage = (int) (endOfPageValueCount - valuesRead);
            if (leftInPage == 0) {
                DataPage page = pageReader.readPage();
                if (page instanceof DataPageV1) {
                    readPageV1((DataPageV1) page);
                } else if (page instanceof DataPageV2) {
                    readPageV2((DataPageV2) page);
                } else {
                    throw new RuntimeException("Unsupported page type: " + page.getClass());
                }
                leftInPage = (int) (endOfPageValueCount - valuesRead);
            }
            int num = Math.min(readNumber, leftInPage);
            if (isCurrentPageDictionaryEncoded) {
                // Read and decode dictionary ids.
                runLenDecoder.readDictionaryIds(
                        num, dictionaryIds, vector, rowId, maxDefLevel, this.dictionaryIdsDecoder);

                if (vector.hasDictionary() || (rowId == 0 && supportLazyDecode())) {
                    // Column vector supports lazy decoding of dictionary values so just set the
                    // dictionary.
                    // We can't do this if rowId != 0 AND the column doesn't have a dictionary (i.e.
                    // some
                    // non-dictionary encoded values have already been added).
                    vector.setDictionary(new ParquetDictionary(dictionary, descriptor));
                } else {
                    readBatchFromDictionaryIds(rowId, num, vector, dictionaryIds);

View on GitHub (pinned to 2f3c205e92)

Solutions

1. Check for parquet-mr version conflicts: mvn dependency:tree / classpath scan; exclude extra parquet-mr from user jars so flink-parquet's version wins
2. Verify the file with an independent reader to rule out corruption; regenerate if broken
3. Align writer, reader, and bundled parquet-mr versions

// detect parquet-mr version conflicts before job submission
ClassLoader cl = Thread.currentThread().getContextClassLoader();
Enumeration<URL> urls = cl.getResources("org/apache/parquet/column/page/DataPageV1.class");
if (Collections.list(urls).size() > 1) throw new IllegalStateException("Multiple parquet-mr jars on classpath");

catch (RuntimeException e) { if (e.getMessage() != null && e.getMessage().startsWith("Unsupported page type")) { /* fix parquet-mr version conflict in user jars */ } else throw e; }

When it happens

Trigger: readToVector hitting endOfPageValueCount == valuesRead and readPage() yielding an unknown DataPage subclass - e.g. a newer parquet-mr version mixed into the user classpath supplying pages the Flink reader does not recognize.

Common situations: User jars bundling a different parquet-mr version than flink-parquet depends on; shaded-dependency leaks in connectors; rarely, corrupted page data parsed into a bogus page object.

Related errors

• could not read page {} in col {}
• CRT transport requested (s3.crt.enabled=true) but the aws-cr
• Only support seek at first.
• A stream against this file was already created.
• Please use AvroParquetReaders.forSpecificRecord(Class<T>) fo

AI-assisted analysis of apache/flink@2f3c205e92 (2026-08-14). Data as JSON: /api/errors/0f49e2d3a5a3d26e. Report an issue: GitHub.