prestodb/presto · error · PrestoException

HIVE_UNKNOWN_ERROR

HIVE_UNKNOWN_ERROR

Error message

HIVE_UNKNOWN_ERROR (message from cause throwable)

What it means

The catch-all branch of propagatePrestoException: any throwable from background split loading that is neither a PrestoException nor a FileNotFoundException is wrapped in HIVE_UNKNOWN_ERROR with the cause as the message source. It indicates an unexpected failure during split production for which no specific error code applies.

Source

Thrown at presto-hive/src/main/java/com/facebook/presto/hive/HiveSplitSource.java:833

            T current = atomicReference.get();
            if (!predicate.test(current)) {
                return false;
            }
            if (atomicReference.compareAndSet(current, newValue)) {
                return true;
            }
        }
    }

    private static RuntimeException propagatePrestoException(Throwable throwable)
    {
        if (throwable instanceof PrestoException) {
            throw (PrestoException) throwable;
        }
        if (throwable instanceof FileNotFoundException) {
            throw new PrestoException(HIVE_FILE_NOT_FOUND, throwable);
        }
        throw new PrestoException(HIVE_UNKNOWN_ERROR, throwable);
    }

    interface PerBucket
    {
        ListenableFuture<?> offer(OptionalInt bucketNumber, InternalHiveSplit split);

        ListenableFuture<List<ConnectorSplit>> borrowBatchAsync(OptionalInt bucketNumber, int maxSize, Function<List<InternalHiveSplit>, BorrowResult<InternalHiveSplit, List<ConnectorSplit>>> function);

        default ListenableFuture<List<ConnectorSplit>> borrowBatchAsync(OptionalInt bucketNumber, Map<String, String> partitionValues, int maxSize, Function<List<InternalHiveSplit>, BorrowResult<InternalHiveSplit, List<ConnectorSplit>>> function)
        {
            throw new UnsupportedOperationException("partition-aware borrowBatchAsync requires a PerBucket implementation that supports partition values");
        }

        void noMoreSplits();

        boolean isFinished(OptionalInt bucketNumber);

        default boolean isFinished(OptionalInt bucketNumber, Map<String, String> partitionValues)

View on GitHub (pinned to 55bb57d202)

Solutions

1. Inspect the underlying cause (getCause()) in coordinator logs to find the real failure.
2. Fix the root cause: metastore connectivity, credentials/tokens, storage timeouts, or network issues.
3. Retry the query if the failure was transient (network blip).
4. If reproducible, capture the cause stack trace and report as a bug with the full exception chain.

boolean isUnknownHiveError(PrestoException e) {
    return HIVE_UNKNOWN_ERROR.toErrorCode().getCode() == e.getErrorCode().getCode();
}

try { runQuery(sql); } catch (PrestoException e) {
    if (isUnknownHiveError(e)) {
        logRootCause(e.getCause());
        if (isTransient(e.getCause())) retryWithBackoff(sql); else throw e;
    } else { throw e; }
}

When it happens

Trigger: Async split enumeration tasks fail with arbitrary exceptions (IOException on the metastore/HDFS client, deserialization errors, network resets, NPEs inside the split source) which are rethrown through propagatePrestoException.

Common situations: Metastore connectivity drops mid-query; HDFS/S3 client timeouts; corrupt file metadata; bugs in storage handler code paths; cluster authentication token expiry.

Related errors

• HIVE_UNKNOWN_ERROR
• partitionHandle must be NOT_PARTITIONED

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/e41ebe10ceaca2c9. Report an issue: GitHub.