Query limits for external data

Limits that apply when a Data Fabric / Data Service query reads from an external system, covering the filter requirement, warehouse paging, result size, and timeouts.

A query over a Federated entity runs against a live system of record rather than against data stored in UiPath. The following limits keep a single query from overwhelming a source system, and keep results repeatable across pages.

They apply only to queries that touch an entity with external fields. A query over data stored entirely in UiPath is not affected by any of them.

A filter is required

A query over an entity that has external fields must include a filter.

The requirement does not apply to:

• a query in which every entity involved is stored entirely in UiPath;
• browsing data in the Data tab of the Data Fabric / Data Service interface;
• a sample read, meaning a single entity with a limit of 200 records or fewer and no offset.

A join condition is not a filter for the purpose of this rule.

Without a filter, the query fails, and the message names the entity that requires one. A filter on any field of that entity satisfies the requirement.

Observação:

The Data tab is exempt, so a filterless query that succeeds while browsing can fail when the same query runs from an activity, an App, or the API.

Data warehouse sources return a single page

Redshift, Snowflake, Databricks, and warehouse databases reached through Database Hub do not enforce primary keys, and they page through results in an order the warehouse does not guarantee. Reading a second page could repeat records or skip them, so a request for a second page is refused.

A filter that narrows the result enough to fit in a single page avoids the limit.

The limit applies to each warehouse source in its own right, including when the warehouse is one of the objects a query reads from. A query that spans more than one object can require more than one page from its warehouse source, and is refused at the point the second page is requested.

Result size

A query whose intermediate result exceeds the processing limit fails, and the message asks for the query to be narrowed. A filter, or a smaller set of selected fields, reduces the result below the limit.

Timeouts and retries

A request to a source system is subject to a time limit, and a query that exceeds it fails with a timeout.

When a source system rate-limits a request, Data Fabric / Data Service retries it with increasing delays. Other rejections by the source system are returned immediately, without a retry.