Troubleshooting - CloudGaming

Overview

This guide covers common issues, their causes, and solutions based on actual implementation details from the CloudGaming source code.

Connection Issues

Client fails to connect to signaling server

Symptoms:

WebSocket connection fails
Browser console shows connection refused
No WebSocket upgrade occurring

Common Causes:

Wrong URL or port

# Check signaling server is running
curl http://localhost:3002/healthz
# Should return: ok

CORS blocking connection

// Server/mm_server/Matchmaker.js:88-92
res.setHeader('Access-Control-Allow-Origin', '*');
res.setHeader('Access-Control-Allow-Methods', 'GET, POST, OPTIONS');

Verify CORS headers are present in response.

WSS required in production

// Server/ScalableSignalingServer.js:199-202
if (config.requireWss && request.headers['x-forwarded-proto'] !== 'https') {
    ws.close(1008, 'WSS required');
}

Use wss:// URL in production, not ws://.

Invalid roomId

// Server/ScalableSignalingServer.js:129-133
const ROOM_ID_REGEX = /^[A-Za-z0-9_\-:.]+$/;
if (typeof roomId !== 'string' || roomId.length > config.roomIdMaxLength) {
    // Connection rejected
}

Room IDs must be alphanumeric with _, -, :, . only.

Solutions:

// Correct client connection
const signalingUrl = 'wss://your-server.com';
const roomId = 'game-room-1';  // Valid format
const ws = new WebSocket(`${signalingUrl}?roomId=${roomId}`);

Logs to Check:

grep "connection" signaling.log
grep "Invalid roomId" signaling.log

ICE connection fails or times out

Symptoms:

ICE connection state stuck in checking
No ICE candidates generated
Connection timeout after 30 seconds

Common Causes:

TURN server not configured

# Check environment variables
echo $PION_TURN_URL
echo $PION_TURN_USERNAME
echo $PION_TURN_CREDENTIAL

Firewall blocking UDP

# Test UDP connectivity to TURN server
nc -u turn.example.com 3478

STUN timeout

// Default STUN server
{ urls: 'stun:stun.l.google.com:19302' }

Try alternative STUN servers if Google STUN is blocked.

Symmetric NAT
- Requires TURN relay
- Host/srflx candidates won’t work

Solutions:

# Configure TURN server
export PION_TURN_URL="turn:turn.example.com:3478"
export PION_TURN_USERNAME="username"
export PION_TURN_CREDENTIAL="password"

# Or use multiple TURN servers
export PION_TURN_URLS="turn:us-west.example.com:3478,turn:us-east.example.com:3478"

Debug ICE candidates:

peerConnection.onicecandidate = (event) => {
    if (event.candidate) {
        console.log('ICE candidate type:', event.candidate.type);
        console.log('ICE candidate:', event.candidate.candidate);
    }
};

Expected types:

host - Direct connection (best)
srflx - Server reflexive via STUN (good)
relay - Via TURN relay (fallback)

Verify ICE gathering:

# Should see candidates in logs
grep "Local ICE Candidate" client.log

Redis connection failures

Symptoms:

Signaling server shows Redis errors
Circuit breaker opens
503 Service Unavailable responses

Common Causes:

Redis server not running
```
redis-cli ping
# Should return: PONG
```

Connection string incorrect

// Check config.js
const config = {
    redisUrl: process.env.REDIS_URL || 'redis://localhost:6379'
};

Circuit breaker triggered

// Server/ScalableSignalingServer.js:68-76
if (redisFailureCount >= config.cbErrorThreshold) {
    redisCircuitOpenUntil = Date.now() + config.cbOpenMs;
}

Circuit opens after threshold failures.

Solutions:

# Check Redis connectivity
redis-cli -h your-redis-host -p 6379 ping

# Check Redis logs
redis-cli info replication

# Monitor Redis connection
curl http://localhost:3002/metrics | grep signaling_redis_up
# Should show: signaling_redis_up 1

Restart with fresh connection:

# Flush any stuck connections
redis-cli CLIENT KILL TYPE normal

# Restart signaling server
node Server/ScalableSignalingServer.js

Check circuit breaker status:

curl http://localhost:3002/metrics | grep circuit_breaker
# signaling_circuit_breaker_open 0  (should be 0)

Audio Issues

No audio playing on client

Symptoms:

Video works but no audio
Audio track present but silent
No audio RTP packets received

Common Causes:

Audio track not added to peer connection

// Verify audio track in SDP
console.log(answer.sdp);
// Should contain: m=audio ...

Opus codec not negotiated

// Check SDP for Opus
// Should have: a=rtpmap:111 opus/48000/2

Audio capture not enabled

// config.json
{
  "audio": {
    "processLoopback": {
      "enabled": true,  // Must be true
      "includeProcessTree": true
    }
  }
}

WASAPI initialization failure
- Audio device not found
- Exclusive mode failed
- Format not supported

Solutions:

# Check audio packets being sent
grep "AUDIO" host.log | grep "Frame"
# Should see:
# [Go/Pion] AUDIO RECEIVE: Frame 2500, size=320 bytes, has_data=true

Verify audio track:

peerConnection.ontrack = (event) => {
    if (event.track.kind === 'audio') {
        console.log('Audio track received:', event.track.id);
        const audioElement = document.getElementById('remoteAudio');
        audioElement.srcObject = event.streams[0];
    }
};

Check audio RTP state:

grep "Audio RTP baseline" host.log
# Should see:
# [Go/Pion] Audio RTP baseline established: PTS=12345 us -> RTP=12345

Debug WASAPI:

{
  "audio": {
    "wasapi": {
      "preferExclusiveMode": false,  // Try shared mode first
      "enforceEventDriven": true,
      "devicePeriodMs": 10.0  // Increase if 5.0 fails
    }
  }
}

Audio latency too high

Symptoms:

Audio lags behind video
Echo or delay in audio
Latency > 100ms

Common Causes:

Large frame size

{
  "audio": {
    "frameSizeMs": 20  // Too large for low latency
  }
}

Buffering enabled

{
  "audio": {
    "latency": {
      "enforceSingleFrameBuffering": false  // Should be true
    }
  }
}

Audio queue congestion

grep "Audio Queue Health" host.log
# If shows WARNING or CRITICAL:
# [Go/Pion] Audio Queue Health [WARNING]: avg=2.4

Solutions:

// Low-latency audio config
{
  "audio": {
    "frameSizeMs": 10,
    "latency": {
      "enforceSingleFrameBuffering": true,
      "targetOneWayLatencyMs": 40,
      "strictLatencyMode": true
    },
    "wasapi": {
      "devicePeriodMs": 5.0
    }
  }
}

Monitor queue health:

grep "Audio Queue Health" host.log
# Target: [GOOD]: avg < 2.0

Check bitrate adaptation:

{
  "audio": {
    "bitrateAdaptation": {
      "enabled": true,
      "minBitrate": 64000  // Allow reduction if congested
    }
  }
}

Audio crackling or distortion

Symptoms:

Popping or clicking sounds
Robotic audio
Intermittent audio gaps

Common Causes:

Buffer underruns
- Frame size too small
- CPU overload
- Packet loss

FEC disabled with high loss

{
  "audio": {
    "enableFec": false,  // Should be true
    "expectedLossPerc": 5
  }
}

Opus complexity too high

{
  "audio": {
    "complexity": 10  // Max complexity, high CPU
  }
}

Solutions:

// Robust audio config
{
  "audio": {
    "frameSizeMs": 10,  // Balance latency/stability
    "complexity": 6,    // Moderate complexity
    "enableFec": true,  // Enable FEC
    "expectedLossPerc": 5,
    "bitrateAdaptation": {
      "enabled": true,
      "fecEnableThreshold": 0.03  // Enable at 3% loss
    }
  }
}

Check packet loss:

grep "WebRTC Stats" host.log
# Look for: loss=X%
# Target: <2% loss

Increase frame size if crackling:

{
  "audio": {
    "frameSizeMs": 20  // More stable but higher latency
  }
}

Video Issues

Video not displaying

Symptoms:

Black screen
Video element not playing
No video track received

Common Causes:

Codec mismatch

// Check SDP for H.264
console.log(offer.sdp);
// Should contain: a=rtpmap:96 H264/90000

Track not attached to video element

peerConnection.ontrack = (event) => {
    if (event.track.kind === 'video') {
        videoElement.srcObject = event.streams[0];
        videoElement.play();  // Must call play()
    }
};

NVENC initialization failed
- NVIDIA GPU not found
- Driver too old
- NVENC not supported on GPU
Capture source not available
- Game process not running
- Window not found
- Desktop capture failed

Solutions:

// Proper video element setup
const videoElement = document.getElementById('remoteVideo');
videoElement.autoplay = true;
videoElement.playsInline = true;
videoElement.muted = true;  // Required for autoplay in some browsers

peerConnection.ontrack = (event) => {
    if (event.track.kind === 'video') {
        console.log('Video track received');
        videoElement.srcObject = event.streams[0];
    }
};

Check video packets:

grep "video" host.log | grep -i "sample\|frame"
# Should see frames being sent

Verify NVENC:

nvidia-smi
# Check GPU is detected and driver version

Debug capture:

{
  "host": {
    "targetProcessName": "YourGame.exe",  // Verify process name
    "window": {
      "resizeClientArea": true,
      "targetWidth": 1920,
      "targetHeight": 1080
    }
  }
}

High video latency

Symptoms:

Glass-to-glass latency > 100ms
Noticeable input lag
Video feels sluggish

Common Causes:

Encoder preset too slow

{
  "video": {
    "preset": "p4"  // p4-p7 are too slow for real-time
  }
}

B-frames enabled

{
  "video": {
    "bf": 2  // B-frames add latency
  }
}

Queue depth too high

{
  "capture": {
    "maxQueueDepth": 4  // Should be 1-2 for low latency
  }
}

Network congestion

grep "RTT" host.log
# High RTT indicates network issues

Solutions:

// Low-latency video config
{
  "video": {
    "preset": "p1",    // Fastest preset
    "rc": "cbr",       // Constant bitrate
    "bf": 0,           // No B-frames
    "rcLookahead": 0,  // No lookahead
    "asyncDepth": 1    // Minimal async pipeline
  },
  "capture": {
    "maxQueueDepth": 1,
    "framePoolBuffers": 2
  }
}

Monitor encoder latency:

grep "encode" host.log | grep -i "latency\|duration"
# Target: <5ms per frame

Check pacer queue:

grep "pacer" host.log
# Queue should stay near 0-1

Poor video quality or artifacts

Symptoms:

Blocky video
Compression artifacts
Blurry motion
Pixelation

Common Causes:

Bitrate too low

{
  "video": {
    "bitrateStart": 4000000  // 4 Mbps too low for 1080p
  }
}

Preset too fast

{
  "video": {
    "preset": "p1"  // Fastest but lowest quality
  }
}

Packet loss

grep "WebRTC Stats" host.log | grep "loss"
# High packet loss degrades quality

PLI (Picture Loss Indication) threshold too low

{
  "video": {
    "minPliLossThreshold": 0.05  // Too aggressive
  }
}

Solutions:

// High-quality video config
{
  "video": {
    "bitrateStart": 10000000,  // 10 Mbps
    "bitrateMin": 8000000,
    "bitrateMax": 15000000,
    "preset": "p3",           // Better quality
    "rc": "vbr_hq",           // High-quality VBR
    "minPliLossThreshold": 0.15  // Less aggressive keyframes
  }
}

Enable FEC for video (if supported):

grep "nack\|pli" host.log
# Monitor retransmission requests

Check bitrate adaptation:

grep "bitrate" host.log
# Verify bitrate is not being reduced too aggressively

Host Registration Issues

Host heartbeat failing

Symptoms:

Host shows as offline in matchmaker
Clients can’t find host
401/403 errors on heartbeat

Common Causes:

Invalid or missing host secret

# Check authorization header
curl -X POST http://localhost:8080/api/host/heartbeat \
  -H "Authorization: Bearer WRONG_SECRET" \
  -H "Content-Type: application/json" \
  -d '{"hostId": "test", "roomId": "room1"}'
# Returns: {"success": false, "error": "Forbidden: Invalid host secret"}

Incorrect heartbeat payload

// Missing required fields
{
  "hostId": "...",  // Required
  "roomId": "..."   // Required
  // region, status, capacity are optional
}

Redis disconnected

# Check matchmaker Redis connection
redis-cli ping

Heartbeat interval too slow

{
  "matchmaker": {
    "heartbeatIntervalMs": 30000  // Should be < 30000 (TTL)
  }
}

Solutions:

# Verify host secret matches
echo $HOST_SECRET
# Should match config.json:

{
  "host": {
    "matchmaker": {
      "hostSecret": "HELLO-MFS",  // Must match
      "heartbeatIntervalMs": 20000  // Send every 20s (TTL is 30s)
    }
  }
}

Test heartbeat manually:

curl -X POST http://localhost:8080/api/host/heartbeat \
  -H "Authorization: Bearer HELLO-MFS" \
  -H "Content-Type: application/json" \
  -d '{
    "hostId": "550e8400-e29b-41d4-a716-446655440000",
    "roomId": "game-room-1",
    "region": "us-west",
    "status": "idle",
    "capacity": 1,
    "availableSlots": 1
  }'
# Should return: {"success": true, "ttl": 30}

Check host TTL:

curl http://localhost:8080/api/hosts/ttl
# Shows remaining TTL for each host

Monitor stale host pruning:

grep "Pruned stale" matchmaker.log
# Stale hosts are pruned every 10s

Clients can't find available hosts

Symptoms:

Match API returns 404
“No hosts available” message
Hosts registered but not discoverable

Common Causes:

Host status not ‘idle’

// Host heartbeat shows:
{
  "status": "busy",      // Should be "idle"
  "availableSlots": 0     // Should be > 0
}

Region mismatch

# Client requests specific region
curl -X POST http://localhost:8080/api/match/find \
  -d '{"region": "eu-west"}'
# But all hosts are in "us-west"

Hosts expired (TTL reached)

# Check host TTLs
curl http://localhost:8080/api/hosts/ttl
# All TTLs should be > 0

Race condition in allocation

grep "Allocation race" matchmaker.log
# Multiple clients trying to allocate same host

Solutions:

# List all hosts
curl http://localhost:8080/api/hosts
# Should show idle hosts with availableSlots > 0

Verify host configuration:

{
  "matchmaker": {
    "heartbeatIntervalMs": 20000,  // < 30s TTL
    "url": "http://localhost:8080",
    "hostSecret": "HELLO-MFS"
  }
}

Test match without region:

curl -X POST http://localhost:8080/api/match/find \
  -H "Content-Type: application/json" \
  -d '{}'
# Should find any available host regardless of region

Check idle_hosts set:

redis-cli SMEMBERS idle_hosts
# Should list host IDs

redis-cli GET host:550e8400-e29b-41d4-a716-446655440000
# Should show host details with availableSlots > 0

Monitor allocation races:

grep -E "Allocation race|Transaction results" matchmaker.log
# Race conditions are retried automatically

Performance Issues

High CPU usage

Symptoms:

CPU usage > 80% sustained
Frame drops
System unresponsive

Common Causes:

Opus complexity too high

{
  "audio": {
    "complexity": 10  // Max CPU usage
  }
}

Small audio frames

{
  "audio": {
    "frameSizeMs": 5  // 200 packets/sec, high overhead
  }
}

Video preset too slow

{
  "video": {
    "preset": "p7"  // Slowest preset
  }
}

GC pressure from buffer churn

grep "Buffer pool" host.log | grep "hit rate"
# Low hit rate = frequent allocations

Solutions:

// CPU-optimized config
{
  "audio": {
    "complexity": 6,      // Balanced
    "frameSizeMs": 10     // Standard frame size
  },
  "video": {
    "preset": "p2",       // Fast preset
    "asyncDepth": 2       // Pipeline encoding
  }
}

Disable unused features:

{
  "capture": {
    "cursor": false,           // No cursor overlay
    "skipUnchanged": true      // Skip duplicate frames
  },
  "video": {
    "gpuTiming": false         // No GPU timing overhead
  }
}

Monitor buffer pool:

grep "Buffer Pool Statistics" host.log
# Target: >95% hit rate

High memory usage

Symptoms:

Memory usage growing over time
Out of memory errors
System swapping

Common Causes:

Buffer pool leaks

grep "Buffer pool" host.log
# Check for anomalies in allocation counts

GC not aggressive enough

// gortc_main/main.go:1301
debug.SetGCPercent(150)  // May hold too much memory

Large video frame buffers

{
  "capture": {
    "copyPoolSize": 6,       // Large pool
    "framePoolBuffers": 4    // Many frame buffers
  }
}

Solutions:

// More aggressive GC
debug.SetGCPercent(100)  // GC at 2x heap instead of 2.5x

// Reduce buffer pools
{
  "capture": {
    "copyPoolSize": 2,
    "framePoolBuffers": 2
  },
  "video": {
    "hwFramePoolSize": 2  // Reduce hardware frame pool
  }
}

Monitor memory:

# Check buffer pool health
grep "Buffer Pool Health" host.log

# Watch memory usage
watch -n 1 'ps aux | grep gortc'

Diagnostic Commands

Quick Health Check

# Signaling server health
curl http://localhost:3002/healthz
curl http://localhost:3002/readyz
curl http://localhost:3002/metrics

# Matchmaker health
curl http://localhost:8080/healthz
curl http://localhost:8080/api/hosts
curl http://localhost:8080/api/hosts/ttl

# Redis health
redis-cli ping
redis-cli INFO stats
redis-cli CLIENT LIST

Log Analysis

# WebRTC stats
grep "WebRTC Stats" host.log | tail -20

# Audio health
grep "Audio Queue Health" host.log | tail -10

# Connection issues
grep -E "error|failed|timeout" signaling.log

# Performance metrics
grep -E "Buffer Pool|latency" host.log | tail -50

Network Diagnostics

# Test TURN connectivity
telnet turn.example.com 3478

# Test STUN
nc -u stun.l.google.com 19302

# Check RTT to signaling server
ping -c 10 your-signaling-server.com

# Monitor bandwidth
iftop -i eth0

Next Steps

Monitoring - Set up comprehensive monitoring
Performance Tuning - Optimize for your use case
GitHub Issues - Report bugs or request features

​Overview

​Connection Issues

​Audio Issues

​Video Issues

​Host Registration Issues

​Performance Issues

​Diagnostic Commands

​Quick Health Check

​Log Analysis

​Network Diagnostics

​Next Steps

Overview

Connection Issues

Audio Issues

Video Issues

Host Registration Issues

Performance Issues

Diagnostic Commands

Quick Health Check

Log Analysis

Network Diagnostics

Next Steps