the request timed out. Try again or use streaming for long r

Error message

the request timed out. Try again or use streaming for long responses

What it means

Exported sentinel ErrTimeout for the LLM app-access proxy: the inference provider request timed out (mapped, for example, from Anthropic timeout/overloaded error types). The message advises streaming because non-streaming requests buffer the whole response and are more prone to hitting the timeout.

Source

Thrown at lib/srv/app/llm/errors/errors.go:31

//
// You should have received a copy of the GNU Affero General Public License
// along with this program.  If not, see <http://www.gnu.org/licenses/>.

package errors

import (
	"errors"
	"fmt"
	"net/http"

	"github.com/gravitational/trace"

	"github.com/gravitational/teleport/lib/utils"
)

var (
	// ErrTimeout returned when the request times out.
	ErrTimeout = errors.New("the request timed out. Try again or use streaming for long responses")
	// ErrBadRequest returned when the request has bad format or invalid fields.
	ErrBadRequest = errors.New("the inference provider rejected the request as invalid. Check the request body for unsupported or invalid fields")
	// ErrCanceled returned when the request is canceled.
	ErrCanceled = errors.New("the request was canceled")
	// ErrUnauthorized returned when the request is unauthorized.
	ErrUnauthorized = errors.New("the inference provider rejected the request due to authentication or authorization configuration. Contact your Teleport administrator")
	// ErrRejected returned when the provider rejects the request.
	ErrRejected = errors.New("the inference provider rejected the request due to usage limits. Contact your Teleport administrator")
	// ErrUnsupported returned when the requested endpoint is not supported.
	ErrUnsupported = errors.New("teleport doesn't support the requested endpoint, please check the list of supported endpoints in the documentation")
	// ErrBadResponse returned when the provider replied the request with an unsupported message or format.
	ErrBadResponse = errors.New("the inference provider returned an unexpected response. Contact your Teleport administrator")
	// ErrConfig returned when the app or app service are misconfigured, requiring admin intervention.
	ErrConfig = errors.New("unable to serve request due to an app configuration error. Contact your Teleport administrator")
	// ErrInternal returned when there is a Teleport processing error (nothing to do with the inference provider).
	ErrInternal = errors.New("unable to serve the request due to an internal error. Contact your Teleport administrator")
	// ErrLimitExceeded returned when Teleport rejects the request due to limit exceeded.
	ErrLimitExceeded = errors.New("tokens quota exceeded. Contact your Teleport administrator")

View on GitHub (pinned to 1283425b60)

Solutions

Retry the request
Switch the client to streaming mode for long responses
If persistent, check provider latency and any configured request timeout limits

Defensive patterns

Strategy: retry

When it happens

Trigger: Thrown at lib/srv/app/llm/errors/errors.go:31 when the library encounters an invalid state.

Common situations: See trigger scenarios.

Understand the failure class

Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

AI-assisted analysis of gravitational/teleport@1283425b60 (2026-09-02). Data as JSON: /api/errors/ca14d55a7d55951c. Report an issue: GitHub.