What does this error mean ? And how can I add a fallback when this happens?

Keynar · May 29, 2026, 9:06pm

I was testing my production voice agent and I got hit with this error : livekit_api::signal_client:530:livekit_api::signal_client - dropping pass-through signal — no stream available

Oh and I also got this when looking in the logs : ERROR:livekit.agents:speech not done in time after interruption, cancelling the speech arbitrarily.

CWilson · May 29, 2026, 9:18pm

Hard to say without the surrounding logs. It comes from here:

github.com/livekit/rust-sdks

livekit-api/src/signal_client/mod.rs

main


      
              // the new stream. Same code path for the steady-state send — the lock is
              // free and we send immediately.
              if let Some(stream) = self.stream.read().await.as_ref() {
                  if let Err(SignalError::SendError) = stream.send(signal.clone()).await {
                      if !pass_through {
                          self.queue.lock().await.push(signal);
                      } else {
                          log::warn!("dropping pass-through signal — send failed");
                      }
                  }
              } else if !pass_through {
                  // Stream not in place AND signal is queueable — hold it.
                  self.queue.lock().await.push(signal);
              } else {
                  log::warn!("dropping pass-through signal — no stream available");
              }
          }
          
          pub async fn flush_queue(&self) {
              let mut queue = self.queue.lock().await;
              if queue.is_empty() {

Muhammad_Usman_Bashir · June 1, 2026, 7:31am

@Keynar, The first warning means a WebRTC negotiation signal (ICE candidate, SDP offer/answer, SyncState, Leave) hit the SDK while the WebSocket was disconnected. These can’t be queued because they only make sense at send time, so the SDK drops them and logs that line [ rust-sdks/livekit-api/src/signal_client/mod.rs:540, with the pass-through list at line 642 ]. Occasional ones during reconnects are normal; a flood means your signaling is unstable and the fix is network-side, not in your code.

The second is the framework’s own safety net firing. On interrupt, the agent sets an interrupt future and schedules a timer; if your TTS or avatar audio doesn’t stop within INTERRUPTION_TIMEOUT, the timer force-cancels them and logs that error [ livekit-agents/livekit/agents/voice/speech_handle.py:220 ]. So the fallback is already running for you. What to fix is whichever component is slow to honor cancellation, usually a TTS that buffers too far ahead. A faster-stopping TTS or tighter first-chunk loop clears it.

Topic		Replies	Views
Livekit API Error Server SDKs agent-sdk-python , python	2	27	June 11, 2026
Signal connection times out on the "v0 path" at agent join, forcing a fallback that adds 0.5–5s of call-setup latency Getting Started	30	117	June 18, 2026
[ISSUE] SIP bridge sends BYE during LLM+TTS processing gap (~14s silence), Twilio Elastic SIP Trunk reports "caller hung up" Telephony agent-development , stt , tts , sip-twilio , sip-trunking , livekit-cloud	9	54	May 12, 2026
Error from STT while in console --text Getting Started	6	70	March 13, 2026
Adaptive interruption disabled due to unrecoverable error, falling back to VAD-based interruption Agents agent-development	4	91	April 14, 2026

What does this error mean ? And how can I add a fallback when this happens?

Related topics