What causes "Error committing write to Hive" in prestodb/presto?

Retry the query after confirming HDFS is healthy — most causes are transient I/O faults.

prestodb/presto · error · PrestoException

HIVE_WRITER_CLOSE_ERROR

HIVE_WRITER_CLOSE_ERROR

Error message

Error committing write to Hive

What it means

RecordFileWriter.commit() closes the underlying Hive RecordWriter to finalize the output file. If closing throws an IOException, the failure is wrapped in HIVE_WRITER_CLOSE_ERROR with this message. The output file could not be finalized and the write should be treated as failed.

Source

Thrown at presto-hive/src/main/java/com/facebook/presto/hive/RecordFileWriter.java:194

        try {
            recordWriter.write(serializer.serialize(row, tableInspector));
        }
        catch (SerDeException | IOException e) {
            throw new PrestoException(HIVE_WRITER_DATA_ERROR, e);
        }
    }

    @Override
    public Optional<Page> commit()
    {
        try {
            recordWriter.close(false);
            committed = true;
            return Optional.empty();
        }
        catch (IOException e) {
            throw new PrestoException(HIVE_WRITER_CLOSE_ERROR, "Error committing write to Hive", e);
        }
    }

    @Override
    public void rollback()
    {
        try {
            try {
                recordWriter.close(true);
            }
            finally {
                // perform explicit deletion here as implementations of RecordWriter.close() often ignore the abort flag.
                path.getFileSystem(conf).delete(path, false);
            }
        }
        catch (IOException e) {
            throw new PrestoException(HIVE_WRITER_CLOSE_ERROR, "Error rolling back write to Hive", e);
        }

View on GitHub (pinned to 55bb57d202)

Solutions

Retry the query after confirming HDFS is healthy — most causes are transient I/O faults.
Inspect the wrapped IOException to find the storage-level problem (quota, DataNode, permissions).
Check remaining HDFS quota and repair the write path if files were left in staging.
If a custom RecordWriter/InputFormat is in use, verify its close() implementation is correct.

Defensive patterns

Strategy: retry

Try / catch

try {
    executeInsert(sql);
} catch (PrestoException e) {
    if (e.getErrorCode() == HiveErrorCode.HIVE_WRITER_CLOSE_ERROR.toErrorCode().getId()
            && e.getCause() instanceof IOException) {
        // verify HDFS health then retry; clean up staging leftovers
    }
}

Prevention

Ensure HDFS quota and disk headroom before large record-file writes.
Avoid concurrent maintenance (node restarts, decommissions) during INSERTs.
Audit custom RecordWriter implementations for correct close() behavior.

When it happens

Trigger: recordWriter.close(false) throwing IOException at the end of an INSERT using a record-based format (TextFile/SequenceFile etc.) — usually an I/O failure flushing final bytes to HDFS.

Common situations: HDFS DataNode loss or network error during the final flush; disk full/quota at commit time; buggy custom RecordWriter implementations that throw on close; long-running writes hitting a cluster change (restart, decommission).

Related errors

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/1dfd1400bcb137b8. Report an issue: GitHub.