Live Metrics
transcode_queue_depth
—
jobs in queue
stream_buffer_ratio
—
CDN buffer utilisation
license_slots_available
—
of 8 slots free
worker_count
—
active transcoding workers
CDN edge →
—
● HEALTHY
Chaos Injection
Demo Timeline — OOM Example
0s
Click "Node OOM Crash"
node-03 drops offline · OOM log emitted
~4s
Grafana alert fires
render_node_up==0 rule triggers webhook
~7s
Gemini queries Grafana MCP
grafana_query_metrics · grafana_query_logs
~11s
Root cause identified
"Isolated OOM · 98% confidence · Fix: restart"
~14s
media_server_restart_node called
node-03 back ONLINE · queue draining
~16s
Annotation + incident written
grafana_create_annotation · cycle complete
Investigator + Remediator Agents (Gemini · ADK)
Ask the Agent