Diagnostics

When a user says "the call was bad", you need numbers to find out why: was it their network, their device, or something else? This page shows how to check a call's health and quality. Use it to:

• show a quality badge in your call screen,
• attach details to a bug report or support ticket,
• spot users who keep having bad calls.

All code runs in your app. You'll see some network words; here's what they mean:

There are two methods:

Both work the same on Web and React Native (it's the same Room class).

getDiagnostics() — quick connection info

const info = room.getDiagnostics();
// {
//   connectionState, reconnectCount, sdkVersion, platform, browser,
//   iceConnectionState, signalingState,
//   remoteIceConnectionState, remotePeerConnectionState,
// }

You can call it any time. It never contains a token or a secret, so you can attach it to a bug report as it is.

The local and remote values can disagree. That's useful: "server says failed, browser says connected" tells you where the problem is.

getConnectionStats() — call quality numbers

const stats = await room.getConnectionStats();
// {
//   connectionState,
//   connectionQuality,      // 'excellent' | 'good' | 'poor' | 'lost' | 'unknown' — the SFU's own read
//   local: TrackStats[],    // one entry per track you've published
//   remote: TrackStats[],   // one entry per track you've subscribed to
// }

Each TrackStats entry looks like this:

{
  kind, direction,          // 'send' | 'receive'
  bitrateBps,                // undefined on a track's very first sample — there's nothing yet to diff against
  packetsLost, packetLossPercent,
  jitterMs,
  roundTripTimeMs,           // only ever present on a send-direction track — WebRTC never reports a receiver's own RTT
  codec,                     // only ever present on a receive-direction video track
  frameWidth, frameHeight, framesPerSecond,
}

This call does real work for every track. Call it every few seconds, not in a tight loop:

setInterval(async () => {
  const stats = await room.getConnectionStats();
  console.log('quality:', stats.connectionQuality);
  for (const track of stats.remote) {
    if (track.packetLossPercent && track.packetLossPercent > 5) {
      console.warn(`${track.kind} losing packets: ${track.packetLossPercent.toFixed(1)}%`);
    }
  }
}, 5000);

On React Native, it's the same call. room is what livqeno.join() returned:

const stats = await room.getConnectionStats();

See stats on the server

You won't always have the user's device in front of you. That's why the same numbers are also sent to Livqeno, where your team can look them up later from a support ticket.

If your SDK version sends stats as telemetry, the numbers show up in Livqeno's Connection records automatically. No extra code needed. View them in the dashboard, or with livqeno connections inspect <id>.

When there are several tracks, they're combined into one number per field:

This works with every client SDK, including Flutter.

Common errors

Tips for production

• Check stats on a timer (every 5 seconds is a good default). Never in a render loop.
• Use connectionQuality as the main answer. It's pushed to you: listen for connectionQualityChanged (or read participant.connectionQuality) rather than polling for it. Only look at the raw TrackStats fields when you need to explain why quality is poor.
• To check a device and network before a call, run the pre-call test.
• Diagnostics never include a token, a secret, or personal data. They're safe to log or send to support.

Next steps

• Reconnection: show a quality indicator using connectionQuality.
• Troubleshooting: what these numbers mean when something's wrong.