prestodb/presto · error · PrestoException

PARQUET_IO_READ_ERROR

PARQUET_IO_READ_ERROR

Error message

Error reading Parquet column 

What it means

LongDecimalFlatBatchReader.readNext wraps any IOException raised while reading a fixed_len_byte_array(16) DECIMAL column chunk (seek, page reads, dictionary page reads) into a PrestoException with code PARQUET_IO_READ_ERROR. It signals that the underlying I/O layer (HDFS, S3, local FS) failed while fetching pages for the decimal column; the original IOException is attached as the cause. This is the reader's generic I/O failure boundary, distinct from decoding-corruption errors.

Source

Thrown at presto-parquet/src/main/java/com/facebook/presto/parquet/batchreader/LongDecimalFlatBatchReader.java:109

        readOffset = readOffset + nextBatchSize;
        nextBatchSize = batchSize;
    }

    @Override
    public ColumnChunk readNext(Optional<DateTimeZone> timezone)
    {
        ColumnChunk columnChunk = null;
        try {
            seek();
            if (field.isRequired()) {
                columnChunk = readWithoutNull();
            }
            else {
                columnChunk = readWithNull();
            }
        }
        catch (IOException exception) {
            throw new PrestoException(PARQUET_IO_READ_ERROR, "Error reading Parquet column " + columnDescriptor, exception);
        }

        readOffset = 0;
        nextBatchSize = 0;
        return columnChunk;
    }

    @Override
    public long getRetainedSizeInBytes()
    {
        return INSTANCE_SIZE +
                (definitionLevelDecoder == null ? 0 : definitionLevelDecoder.getRetainedSizeInBytes()) +
                (valuesDecoder == null ? 0 : valuesDecoder.getRetainedSizeInBytes()) +
                (dictionary == null ? 0 : dictionary.getRetainedSizeInBytes()) +
                (pageReader == null ? 0 : pageReader.getRetainedSizeInBytes());
    }

    protected boolean readNextPage()

View on GitHub (pinned to 55bb57d202)

Solutions

1. Inspect the caused-by IOException to identify the storage layer failure (timeout, missing block, permission).
2. Retry the query — transient S3/HDFS failures often resolve; enable Presto/Hive connector retry settings for splits.
3. Verify the file still exists and is readable by the Presto worker user (permissions, hdfs dfs -test, aws s3 head-object).
4. Check storage health: DataNode status in HDFS, S3 service status/throttling limits.
5. If failures are persistent, restore the file from backup or re-register the table location.

Example fix

// before: hard failure on transient I/O
ColumnChunk chunk = reader.readNext(timezone);

// after: retry transient I/O at the connector level
for (int attempt = 0; attempt < 3; attempt++) {
    try { return reader.readNext(timezone); }
    catch (PrestoException e) {
        if (e.getErrorCode() != PARQUET_IO_READ_ERROR.toErrorCode() || attempt == 2) throw e;
        sleep(backoff(attempt));
    }
}

if (!fs.exists(path) || !fs.canRead(path)) {
    throw new IOException("File missing or unreadable before scan: " + path);
}

try {
    return reader.readNext(timezone);
} catch (PrestoException e) {
    if (PARQUET_IO_READ_ERROR.toErrorCode().equals(e.getErrorCode()) && isRetryable(e.getCause())) {
        return retryWithBackoff(() -> reopenAndRead(), 3);
    }
    throw e;
}

When it happens

Trigger: readNext() on a DECIMAL(38,x) column when the underlying PageReader/seek()/readPage()/readDictionaryPage() throws IOException — e.g. the HDFS block is unavailable, the S3 object read times out, or the file was deleted/renamed mid-scan.

Common situations: S3 throttling (503 SlowDown) or expired credentials during a long scan, HDFS DataNode unavailability or block re-replication, files removed by a retention job while a query is running, network partitions between coordinator and storage, or permission changes on the file.

Related errors

• PARQUET_IO_READ_ERROR
• PARQUET_IO_READ_ERROR
• PARQUET_IO_READ_ERROR
• PARQUET_IO_READ_ERROR
• HUDI_CANNOT_OPEN_SPLIT

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/d6bd100e8782044d. Report an issue: GitHub.