Pipeline Health
Live health of the v11 triage pipeline — connectors → Redis Streams → normalization → pre-filters → correlation → enrichment → RAG → AI Gateway reasoning → analyst console → ticketing. Degraded components fall back to deterministic rules; cases never fail silently.
Active components
1/2
running
System health
91%
weighted avg
Tasks processed
2,612
last 24h
Error rate
0.42%
11 errors
Pipeline status
DEGRADED
1 component below SLO
Pipeline components (2)Per-component health, throughput, latency and uptime.
Correlation Engine
Cross-source pivot graph & case formation · v2.4.1
Running
Health
100%
Processed
1,361
Errors
0
Avg
141ms
Uptime
105d
Last
Just now
AI Gateway · LLM Triage Advisor
Suggested severity (with reasoning), summary, recommended next steps · llama-3.1-8b · v1.2.4
Degraded
Health
82%
Processed
1,251
Errors
11
Avg
1820ms
Uptime
105d
Last
12 sec ago
Task processing volume (7d)Total triage tasks completed across all agents per day.
Average response time (ms, 7d)Per-agent p50 response latency. The LLM Advisor runs at ~10× the deterministic agents — expected.
Agent health is observed from the AI runtime queue (Redis Streams) and vLLM inference logs. Degraded agents auto-fall back to deterministic rules; cases never fail silently.