Apache Iceberg Nodes

Append rows to Apache Iceberg tables on S3 or Google Cloud Storage, one atomic snapshot per execution, committed through an Iceberg REST catalog or AWS Glue.

Needs

An Apache Iceberg connection

Operations

Append to Table

Limits

Appends only; batch rows with a Buffer node

Apache Iceberg nodes write pipeline rows into Iceberg tables that Spark, Trino, Athena, Snowflake, BigQuery and DuckDB read through the same catalog. They connect through your configured Apache Iceberg connections.

Configuration Quick Reference

Field	What you choose	Details
Parameters	Connection, Function, Function Parameters, Timeout Override	Select the connection profile, function, configure function parameters with expression support, and optionally override timeout.
Settings	Description, Timeout (seconds), Retry on Timeout, Retry on Fail, On Error	Node description, maximum execution time, retry behavior on timeout or failure, and error handling strategy. All execution settings default to pipeline-level values.

---

Table Operations

Apache Iceberg Append to Table Node

Zoom

Apache Iceberg Append to Table Node

Write pipeline rows into an Iceberg table as one snapshot, committed through the catalog. The rows of an execution land together or not at all, and are queryable from every engine as soon as the node finishes. The first write creates the table.

Supported Function Types:

Function Name	Purpose	Common Use Cases
Append to Table	Append rows to an Iceberg table	Streaming UNS readings into a lakehouse table queried from Trino or Athena

Node Configuration

Parameter	Type	Required	Description
Connection	Selection	Yes	Apache Iceberg connection profile to use
Function	Selection	Yes	Append to Table function from the selected connection
Function Parameters	Dynamic	Varies	Auto-populated from the function schema (namespace, table, data, columns, partitionColumn, partitionBy). See your Apache Iceberg connection functions for full parameter details.
Timeout Override	Number (seconds)	No	Override the default function timeout

data takes one row (an object) or a batch (an array of objects). Bind it to the rows an upstream node produces, for example {{ $node["Buffer"].result }}.

Batch rows before the node

Every execution writes a data file, a manifest, a manifest list and a new metadata file, and adds a snapshot. Put a Buffer node in front so each write carries hundreds of rows, not one, and expire old snapshots in your catalog or engine.

Output Structure

{
  "result": {
    "table": "plant_ops.sensor_readings",
    "committed": true,
    "created": false,
    "snapshotId": "4417203518092265113",
    "sequenceNumber": 43,
    "rowsWritten": 500,
    "dataFiles": 1,
    "bytesWritten": 18432,
    "parseFailures": 0,
    "snapshots": 43
  },
  "_metadata": { "success": true, "functionId": "<function-id>", "durationMs": 820, "timestamp": "2026-11-02T08:00:00Z" }
}

Field	Type	Description
result.table	string	The table's namespace and name
result.committed	boolean	Whether rows were committed; false only when the input carried no rows
result.created	boolean	Whether this execution created the table (when committed)
result.snapshotId	string	The snapshot this execution committed, as text (Iceberg snapshot IDs are larger than a JSON number keeps exactly); on a replay, the snapshot that already holds its rows (when committed)
result.sequenceNumber	number	The sequence number of snapshotId (when committed)
result.rowsWritten	number	Rows written; 0 on a replay
result.dataFiles	number	Data files this execution added (when committed, and not on a replay)
result.bytesWritten	number	Size of the data files this execution added (when committed, and not on a replay)
result.parseFailures	number	Values that did not parse as their column's type and were written as null. The first failure in each column of a table is also logged as a warning
result.snapshots	number	Snapshots the table holds after this execution (when committed). Each execution adds one; expire old snapshots in your catalog or engine
result.alreadyCommitted	boolean	Present and true when this execution is a replay of a write the table already holds: nothing was written
result.ignoredFields	array	Row fields left out because no column matches them; present only when there are some. The first time each field is left out is also logged as a warning
result.warning	string	Present once the table holds more than 10,000 snapshots, saying to expire them and batch rows

A replayed execution lands once

When store-and-forward, Retry on Fail or a dead-letter replay sends a write again, the table may already hold its rows. The connector recognises that from the write's key, recorded in the snapshot summary as maestrohub.write-key, and answers with alreadyCommitted: true instead of appending a second copy, for as long as that snapshot is kept. An execution whose commit got no answer fails as retryable; its retry finds the key if the commit had landed.

---

Settings Tab

All Apache Iceberg node types share the same Settings tab:

Setting	Type	Default	Description
Description	Text	-	Optional description displayed on the node
Timeout (seconds)	Number	Pipeline default	Maximum time the node may run before timing out
Retry on Timeout	Toggle	Pipeline default	Automatically retry the node if it times out
Retry on Fail	Toggle	Pipeline default	Automatically retry the node if it fails
On Error	Selection	Pipeline default	Error strategy: Pipeline Default (the pipeline's Error Handling setting), Stop Pipeline or Continue Execution

When left at their defaults, these settings inherit from the pipeline-level execution configuration.