prestodb/presto · warning · UncheckedIOException

java.io.IOException (wrapped in UncheckedIOException)

Error message

java.io.IOException (wrapped in UncheckedIOException)

What it means

MapDirectBatchStreamReader.close wraps IOException from closing the nested key and value stream readers in UncheckedIOException. It is a teardown-time failure indicating the underlying ORC streams could not be closed cleanly — typically because the datasource was already closed or a storage error occurred during cleanup.

Source

Thrown at presto-orc/src/main/java/com/facebook/presto/orc/reader/MapDirectBatchStreamReader.java:285

    }

    @Override
    public String toString()
    {
        return toStringHelper(this)
                .addValue(streamDescriptor)
                .toString();
    }

    @Override
    public void close()
    {
        try (Closer closer = Closer.create()) {
            closer.register(keyStreamReader::close);
            closer.register(valueStreamReader::close);
        }
        catch (IOException e) {
            throw new UncheckedIOException(e);
        }
    }

    @Override
    public long getRetainedSizeInBytes()
    {
        return INSTANCE_SIZE + keyStreamReader.getRetainedSizeInBytes() + valueStreamReader.getRetainedSizeInBytes();
    }
}

View on GitHub (pinned to 55bb57d202)

Solutions

Close readers before the OrcDataSource; use a Closer to enforce ordering.
Ensure close() is idempotent in custom streams.
Inspect the wrapped IOException cause; treat as secondary to the original query failure.
Upgrade Presto if this matches a known double-close bug.

Example fix

// before
dataSource.close();
mapReader.close();
// after
try (Closer closer = Closer.create()) {
    closer.register(mapReader::close);
    closer.register(dataSource::close);
}

Defensive patterns

Strategy: try-catch

Try / catch

try {
    mapReader.close();
}
catch (UncheckedIOException e) {
    LOG.warn(e.getCause(), "Failed closing map key/value stream readers");
}

Prevention

Establish close order: readers, then datasource.
Use Closer so key and value readers both close even on failure.
Never close the shared OrcDataSource while readers are active.
Make custom streams tolerate repeated close().

When it happens

Trigger: Calling close() on the map reader after the OrcDataSource was closed; disk/network errors while closing the key or value column streams.

Common situations: Query cancellation closing the source before readers; double-close paths in custom page-source code; transient storage failures (HDFS/S3) at teardown.

Related errors

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/a0d372e9c0120902. Report an issue: GitHub.