test-stream

name: test-stream description: Boot the app, stream via RTMP, watch via HLS/DASH, and validate the full pipeline disable-model-invocation: true

Full Pipeline E2E Tester

This skill walks through a complete end-to-end test of Theatrum: build, start, stream via RTMP, validate HLS output, check metrics, and clean up. Use $ARGUMENTS to customize (e.g., /test-stream passthrough or /test-stream with-recording).

Current Config

!cat config.yml 2>/dev/null || echo "No config.yml found — will create a test config"

Step 1: Prerequisites Check

Verify required tools are installed:

ffmpeg -version | head -1
go version
curl --version | head -1

If any are missing, stop and inform the user.

Step 2: Test Video

If no test input video exists, generate a 30-second test pattern:

ffmpeg -f lavfi -i "testsrc=duration=30:size=1280x720:rate=30" \
       -f lavfi -i "sine=frequency=440:duration=30" \
       -c:v libx264 -preset ultrafast -c:a aac -b:a 128k \
       -pix_fmt yuv420p /tmp/theatrum_test_input.mp4

Step 3: Config Setup

If no config.yml exists, create a minimal test config. Use a passthrough live stream for simplicity:

application:
  public_path: "http://localhost:8080"

server:
  http: 8080
  rtmp: 1935
  rtmp_config:
    reconnect_delay: 5
    cleanup_delay: 10

channels:
  "/test/{username}":
    stream:
      type: live
      path: "test/{username}"
      live_stream_key: "testkey123"
      auth_token_template: "{username}"
      distribution:
        hls:
          segment_duration: 2
          window_size: 3
      viewers:
        enabled: true
        window: 5
      views:
        enabled: true
        window: 5

Step 4: Build & Start

# Build
cd src && go build -o ../theatrum ./cmd/main.go && cd ..

# Start in background
./theatrum &
THEATRUM_PID=$!
echo "Theatrum started with PID $THEATRUM_PID"

# Wait for startup
sleep 2
curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/metrics

Expected: HTTP 200 from /metrics.

Step 5: Compute Auth Token

The RTMP auth token is computed via XOR. For the test config above (key=testkey123, username=alice):

python3 -c "
key = 'testkey123'
username = 'alice'
result = bytes([ord(c) ^ ord(key[i % len(key)]) for i, c in enumerate(username)])
print(result.hex())
"

This outputs the hex token to use as the RTMP publishing name.

Step 6: Stream via RTMP

TOKEN=$(python3 -c "
key = 'testkey123'
username = 'alice'
result = bytes([ord(c) ^ ord(key[i % len(key)]) for i, c in enumerate(username)])
print(result.hex())
")

ffmpeg -re -i /tmp/theatrum_test_input.mp4 \
       -c copy -f flv \
       "rtmp://localhost/test/alice/$TOKEN" &
FFMPEG_PID=$!
echo "FFmpeg streaming with PID $FFMPEG_PID"

Wait a few seconds for segments to be generated:

sleep 8

Step 7: Validate Output

Check that playlists/manifests and segments are being served:

HLS (passthrough)

echo "=== HLS Playlist ==="
curl -s http://localhost:8080/test/alice/default/playlist.m3u8

echo ""
echo "=== Segment check ==="
SEGMENT=$(curl -s http://localhost:8080/test/alice/default/playlist.m3u8 | grep '.ts' | head -1)
if [ -n "$SEGMENT" ]; then
    HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" "http://localhost:8080/test/alice/default/$SEGMENT")
    echo "Segment HTTP status: $HTTP_CODE"
else
    echo "ERROR: No segments found in playlist"
fi

HLS (multi-quality)

curl -s http://localhost:8080/test/alice/master.m3u8
curl -s http://localhost:8080/test/alice/low/playlist.m3u8

DASH

If the channel uses DASH distribution, check the MPD manifest:

echo "=== DASH Manifest ==="
curl -s http://localhost:8080/test/alice/manifest.mpd

echo ""
echo "=== DASH Segment check ==="
# Check an init segment
HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" "http://localhost:8080/test/alice/init-stream0.m4s")
echo "Init segment HTTP status: $HTTP_CODE"

Dual mode (HLS + DASH)

Both endpoints should work:

curl -s -o /dev/null -w "HLS: %{http_code}\n" http://localhost:8080/test/alice/master.m3u8
curl -s -o /dev/null -w "DASH: %{http_code}\n" http://localhost:8080/test/alice/manifest.mpd

Step 8: Check Metrics

echo "=== Theatrum Metrics ==="
curl -s http://localhost:8080/metrics | grep "theatrum_"

Expected metrics to see:

• theatrum_stream_viewers — Concurrent viewers per stream
• theatrum_stream_views — Total views per stream
• theatrum_http_requests_total — HTTP request count
• theatrum_stream_duration_seconds — Active stream durations
• theatrum_ffmpeg_exits_total — FFmpeg process exits
• theatrum_rtmp_received_bytes_total — Bytes received via RTMP

Step 9: Viewer/View Tracking (Optional)

If viewers/views are enabled, test them:

# Simulate a viewer by requesting segments repeatedly
for i in $(seq 1 10); do
    SEGMENT=$(curl -s http://localhost:8080/test/alice/default/playlist.m3u8 | grep '.ts' | tail -1)
    [ -n "$SEGMENT" ] && curl -s -o /dev/null "http://localhost:8080/test/alice/default/$SEGMENT"
    sleep 2
done &

# After the window period, check counts
sleep 10
echo "Viewers: $(curl -s http://localhost:8080/test/alice/viewers.txt)"
echo "Views: $(curl -s http://localhost:8080/test/alice/views.txt)"

Step 10: Cleanup

# Stop FFmpeg
kill $FFMPEG_PID 2>/dev/null

# Wait for stream cleanup
sleep 5

# Stop Theatrum
kill $THEATRUM_PID 2>/dev/null

# Remove test data
rm -rf data/test/
rm -f /tmp/theatrum_test_input.mp4

echo "Cleanup complete"

Troubleshooting