---
title: "Shutdown and Draining"
description: "Learn more about the default behavior of our SDK if the application shuts down unexpectedly."
url: https://docs.sentry.io/platforms/go/guides/negroni/configuration/draining/
---

# Shutdown and Draining | Sentry for Negroni

By default, the SDK uses an asynchronous transport running in its own separate goroutine to send events over the network. As a result, some events may be lost if the application shuts down unexpectedly. The SDK provides mechanisms to handle this scenario.

To avoid unintentionally dropping events when the program terminates, arrange for `sentry.Flush` to be called, typically using `defer`.

If you use multiple clients, arrange for each of them to be flushed as appropriate.

`Flush` waits until any buffered events are sent to the Sentry server, blocking for at most the given timeout. It returns `false` if the timeout was reached. In that case, some events may not have been sent.

```go
// err := sentry.Init(...)
defer sentry.Flush(2 * time.Second)

sentry.CaptureMessage("my message")
```

`FlushWithContext` is an alternative to Flush that uses a context.Context to manage the timeout or cancellation of the flush operation. This approach is particularly useful in applications where you are already using contexts for managing deadlines and cancellations.

```go
// err := sentry.Init(...)
ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
defer cancel()

sentry.CaptureMessage("my message")
sentry.FlushWithContext(ctx)
```