---
id: flow--data-sampler
title: "Data sampler"
summary: "Generates test data: for every message on `Input` it creates a new record of the output connection's type and fills its properties with (optionally seeded) random values. Use it to exercise a flow without a real data source, for example Timer → Data sampler → the flow under test."
audience: user
keywords: "input, output, seed, nullprobability, numbermin, numbermax, collectionminitems, collectionmaxitems"
---
# Data sampler

Generates test data: for every message on `Input` it creates a new record of the output connection's type and fills its properties with (optionally seeded) random values. Use it to exercise a flow without a real data source, for example Timer → Data sampler → the flow under test.

## Pins

| Pin | Kind | Types | Description |
|---|---|---|---|
| Input | input | any | Trigger. Each message produces one sample; the payload is ignored. |
| Output | output | the type of the outgoing connection | A new record of that type filled with random values. Set the type on the outgoing connection; it decides what is generated. |

## Configuration

| Field | Type | Default | Description |
|---|---|---|---|
| `Seed` | string | | Seed text. The same seed and the same (sequential) input sequence produce the same data. Empty = different data every run. |
| `NullProbability` | integer (0-100) | `10` | Percent chance that a nullable property (reference types such as `string`, and `Nullable<T>`) is set to `null`. Properties marked required in the type are never null. |
| `NumberMin` | number | `0` | Lower bound for numeric properties. |
| `NumberMax` | number | `10000` | Upper bound for numeric properties. |
| `CollectionMinItems` | integer (0-10000) | `0` | Minimum number of items in generated lists, arrays and dictionaries. |
| `CollectionMaxItems` | integer (0-10000) | `5` | Maximum number of items in generated lists, arrays and dictionaries. |

## Behaviour

- On start the generator for the output type is built once. If the output pin has no connection/type, nothing is generated and input messages are ignored.
- Generated values:
  - `string`: 8-32 random letters and digits; `char`: one letter or digit; `bool`: random.
  - numbers (`byte` ... `ulong`, `float`, `double`, `decimal`): between `NumberMin` and `NumberMax`, clamped to the range of the property type.
  - `DateTime`, `DateTimeOffset` (UTC), `DateOnly`: between 1970-01-01 and 2100-01-01; `TimeOnly`, `TimeSpan`: within one day.
  - `Guid` (version 4), `IPAddress` (IPv4), enums (one of the defined values), `byte[]`.
  - lists, arrays and dictionaries (string or value-type keys) with `CollectionMinItems`-`CollectionMaxItems` items; duplicate dictionary keys are dropped.
  - nested complex types are filled recursively up to 4 levels deep; deeper levels are `null`. The root record is never `null`.
- Constraints declared on flow type properties are respected: minimum / maximum length for strings and `byte[]`, minimum / maximum value for numbers (the declared range wins over `NumberMin` / `NumberMax` if they do not overlap). Regex validators are not honoured (a warning is shown).
- Unsupported property types (`object`, interfaces, abstract classes, streams, types without a parameterless constructor) are left at their default value with a warning. Read-only and system properties are skipped.
- The metadata of the input message is passed through to the output.

## Notes

- `NumberMin` must not be greater than `NumberMax`, `CollectionMinItems` not greater than `CollectionMaxItems`; otherwise the configuration is rejected. A numeric range that contains no valid value for a property type (for example `NumberMax` = `-1` for a `byte` property) makes the flow fail to start.
- Reproducibility with `Seed` holds only for sequential input; parallel callers (Fork, parallel ForEach) interleave the random sequence.

## Example

```json
{ "Seed": "test-1", "NullProbability": 0, "NumberMin": 1, "NumberMax": 100, "CollectionMinItems": 1, "CollectionMaxItems": 3 }
```

Timer → **Data sampler** (output connection type `MyFlow.Order`) → Data mapper → ...