What causes "Unexpected partitioning: " in prestodb/presto?

Log/inspect the partitioning handle value in the message to see which unsupported scheme was produced

prestodb/presto · error · IllegalArgumentException

Unexpected partitioning:

Error message

Unexpected partitioning:

What it means

Thrown from AbstractPrestoSparkQueryExecution.createPartitioner() when the plan's Partitioning is not one of the supported Spark partitioning schemes (e.g. HASH/SINGLE/FIXED_BROADCAST). This is a programming/invariant error: the query execution received a partitioning handle the Spark execution engine cannot map to a PrestoSparkPartitioner.

Source

Thrown at presto-spark-base/src/main/java/com/facebook/presto/spark/execution/AbstractPrestoSparkQueryExecution.java:337

    protected static Partitioner createPartitioner(PartitioningScheme partitioningScheme)
    {
        PartitioningHandle partitioning = partitioningScheme.getPartitioning().getHandle();
        if (partitioning.equals(SINGLE_DISTRIBUTION)) {
            return new PrestoSparkPartitioner(1);
        }
        if (partitioning.equals(FIXED_HASH_DISTRIBUTION)
                || partitioning.equals(FIXED_ARBITRARY_DISTRIBUTION)
                || partitioning.getConnectorId().isPresent()) {
            int[] bucketToPartition = partitioningScheme.getBucketToPartition().orElseThrow(
                    () -> new IllegalArgumentException("bucketToPartition is expected to be assigned at this point"));
            checkArgument(bucketToPartition.length > 0, "bucketToPartition is expected to be non empty");
            int numberOfPartitions = IntStream.of(bucketToPartition)
                    .max()
                    .getAsInt() + 1;
            return new PrestoSparkPartitioner(numberOfPartitions);
        }
        throw new IllegalArgumentException("Unexpected partitioning: " + partitioning);
    }

    @Override
    public List<List<Object>> execute()
    {
        List<Tuple2<MutablePartitionId, PrestoSparkSerializedPage>> rddResults;
        try {
            tuneMaxExecutorsCount();
            rddResults = doExecute();
            queryStateTimer.beginFinishing();
            PrestoSparkTransactionUtils.commit(session, transactionManager);
            queryStateTimer.endQuery();
        }
        catch (Throwable executionException) {
            queryStateTimer.beginFinishing();
            try {
                PrestoSparkTransactionUtils.rollback(session, transactionManager);
            }

View on GitHub (pinned to 55bb57d202)

Solutions

Log/inspect the partitioning handle value in the message to see which unsupported scheme was produced
Ensure the query plan is produced by the standard Presto-on-Spark planner without custom partitioning extensions
Upgrade coordinator/worker jars so planner and Spark execution agree on supported partitioning types
If adding a new partitioning type, extend createPartitioner to handle it

Example fix

// before
throw new IllegalArgumentException("Unexpected partitioning: " + partitioning);
// after: handle the scheme explicitly before the throw, e.g.
if (partitioning.getHandle().equals(HASH_DISTRIBUTED)) {
    return new PrestoSparkPartitioner(numberOfPartitions);
}
throw new IllegalArgumentException("Unexpected partitioning: " + partitioning);

Defensive patterns

Strategy: validation

Validate before calling

Partitioning p = ...;
Set<PartitioningHandle> supported = ImmutableSet.of(HASH_DISTRIBUTED.getHandle(), SINGLE_DISTRIBUTION.getHandle(), FIXED_BROADCAST_DISTRIBUTION.getHandle());
if (!supported.contains(p.getHandle())) throw new IllegalArgumentException("unsupported: " + p);

Type guard

boolean isSupportedPartitioning(Partitioning partitioning) {
    PartitioningHandle h = partitioning.getHandle();
    return h.equals(HASH_DISTRIBUTED.getHandle()) || h.equals(SINGLE_DISTRIBUTION.getHandle()) || h.equals(FIXED_BROADCAST_DISTRIBUTION.getHandle());
}

Try / catch

try {
    Partitioner partitioner = execution.partitioner();
}
catch (IllegalArgumentException e) {
    if (e.getMessage().startsWith("Unexpected partitioning:")) {
        // fail fast with plan dump for planner debugging
    }
    throw e;
}

Prevention

Avoid custom planner extensions that emit novel partitioning handles
Keep planner and Spark execution modules on the same version
When adding partitioning types, update createPartitioner in the same change
Log the partitioning handle during plan validation

When it happens

Trigger: partitioner() is called with a Partitioning handle that does not match any expected case, so control reaches the final throw new IllegalArgumentException("Unexpected partitioning: " + partitioning).

Common situations: Custom connectors or planner extensions emitting an unsupported partitioning; version mismatches where a new partitioning type is produced upstream but the Spark execution path was not updated.

Related errors

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/f3001fbeac32f3fb. Report an issue: GitHub.