> ## Documentation Index
> Fetch the complete documentation index at: https://docs.021labs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# flush()

> Manually flush pending traces

## Signature

```python theme={null}
def flush() -> None
```

## Parameters

None.

## Example

```python theme={null}
from rdk import init, flush

init()

# Your LLM calls...

# Force send all pending traces
flush()
```

## Behavior

When called, `flush()`:

1. Takes all spans from the internal queue
2. Groups them by trace ID
3. Sends them to the collector immediately
4. Does not wait for the batch size or flush interval

## Use Cases

**Debugging**: Force traces to appear in your dashboard immediately.

```python theme={null}
@app.post("/debug-chat")
@observe(name="debug")
async def debug_chat(request: Request):
    result = await chat(request.message)
    flush()  # See trace immediately
    return result
```

**Before exit**: Ensure all traces are sent before the process exits.

```python theme={null}
import atexit
from rdk import flush

atexit.register(flush)
```

**Manual control**: Disable automatic flushing and control timing.

```python theme={null}
init(flush_interval=999999)  # Effectively disable auto-flush

# Flush when you decide
flush()
```

## Thread Safety

`flush()` is thread-safe and can be called from any thread.

## See Also

* [shutdown()](/api-reference/shutdown) - Flush and cleanup
* [init()](/api-reference/init) - Configure flush interval