Technical Limits

This page describes the known limitations and constraints that apply when using database nodes in Integration Flow.

Query Result Preview Limit

When testing a query directly in the node configuration (using the Test Query feature), the results preview is limited to 25 rows. This is a UI constraint only — the actual query executed at runtime returns all rows the query produces.

Temporary Tables

Temporary tables (created using CREATE TEMP TABLE) are supported in Flow, but come with two important constraints you need to be aware of before using them.

Use Single-Row Nodes Only

A temporary table exists only within the database connection that created it — it is invisible to any other connection. Because of this, every node that interacts with a temporary table must be a Single-Row node.

This is because Single-Row nodes reuse the same database connection throughout the entire flow. Multi-row nodes do not have this guarantee — they may open a different connection, and since the temporary table does not exist on that connection, the query will fail.

Temporary Tables Are Not Visible Across Flow Boundaries

Each flow — whether a main flow or a subflow — operates on its own database connection. Session-scoped objects like temporary tables are tied to that connection and are not shared.

In practice this means:

• A temporary table created in the main flow cannot be read or written by a subflow.

• A temporary table created in a subflow is gone by the time the subflow returns — the main flow cannot access it.

This applies equally in both directions. The flows simply do not share a connection, so they cannot see each other's temporary state.

If you need to share data between a main flow and a subflow, pass it through flow input/output variables, or persist it to a regular (permanent) table and read it from there.

Integration Flow instead of Decision Flow

Database nodes are available in both Decision Flow and Integration Flow, but the two differ significantly in how long they can run — and that difference matters a lot for database operations.

Decision Flow has a hard execution limit of 1–8 seconds (depending on your plan or on-premise configuration). Any database query that takes longer than the remaining time will cause the entire flow to fail with a timeout. This makes Decision Flow suitable only for fast, lightweight queries — simple lookups or single-row fetches where you can guarantee a short response time.

Integration Flow was built for longer-running work. The flow itself can run for up to 24 hours, and database nodes can stay active as long as there is ongoing activity — though any single query that produces no activity for 5 minutes will be terminated.

See Decision Flow vs. Integration Flow for a full comparison.