For AWS-based pipelines (MediaLive, MediaPackage, IVS), navigate to CloudWatch Metrics and explore the relevant service namespace (e.g., AWS/IVS or AWS/MediaLive) to find stream health metrics such as IngestVideoBitrate, IngestAudioBitrate, and DroppedFrames.
Create CloudWatch Alarms on the DroppedFrames metric with a threshold of > 0 over a short evaluation period (e.g., 1 minute) to alert on any frame loss during the stream.
Set an alarm on IngestVideoBitrate falling below a minimum expected value to detect encoder issues or network degradation on the contribution path.
For platform-agnostic monitoring, configure the encoder to push stats to a time-series database (e.g., InfluxDB or Prometheus) via encoder status APIs or OBS websocket; build dashboards showing bitrate, FPS, and RTT.
Use Mux Data or a video observability platform to correlate ingest health with viewer-side rebuffering rate and stream error events from the player SDK.
Set up SNS notifications on critical alarms to page on-call engineers when stream health degrades below acceptable thresholds.
Known gotchas
DroppedFrames in CloudWatch for IVS measures frames dropped at the ingest point, not at the player; player buffering issues require separate monitoring via the player SDK.
Metric granularity varies by service; some MediaLive metrics are only available at 1-minute granularity, which may be too coarse for rapid fault detection on live events.
High-velocity metrics (multiple streams) can generate significant CloudWatch costs; scope alarms to active streams only and use metric math to avoid unnecessary alarm evaluations.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp