<#C088ZNU7QQ5> i observed high latency in one of m...
# ask-ai
s
#C088ZNU7QQ5 i observed high latency in one of my session and observed on agent track is this related to any recent outage what to make out of this graph specially the connection latency graph
✅ 1
t
Thanks for reaching out, shekhar. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
Thanks — I looked at the Connection Latency graph and your session details (RM_xT2rroIMwYV3, EG_0xSNn4zEQbNg). I’ll address whether this looks like an outage and how to interpret the graph. What the graph shows • The chart plots percentile latency (p50 / p90 / p99). Spikes around 1.5–2.0 seconds mean a large portion of connections experienced round‑trip delays in that range (not milliseconds). • Latency at 1.5–2s is high for real‑time voice (perceptible lag). Typical LiveKit/WebRTC audio transport is expected to be much lower. Is this likely a recent outage? • This is unlikely to be a platform‑wide outage. The session was routed and completed (egress recorded to AWS), and LiveKit’s infrastructure is resilient. The observed pattern points to localized network or client issues rather than a service outage. What to investigate next (quick checklist) 1. Client network & device ◦ Check Wi‑Fi/cellular quality, packet loss, and whether client was behind a congested NAT, VPN, or restrictive firewall. 1. Region/routing ◦ Confirm the client was connected to the nearest LiveKit region/edge. 1. TURN vs direct ◦ Determine whether the client used a TURN relay (can increase latency) versus a direct peer path. 1. Run a connectivity check ◦ Use the ConnectionCheck/diagnostic utility from the client environment to test WebSocket, WebRTC, and TURN paths. 1. Correlate telemetry ◦ Compare the connection latency spikes to per‑participant connection quality, bandwidth charts, and any client-side logs/timestamps to find where latency began. If you want, share: • Client network type (Wi‑Fi/cellular), device OS/browser, approximate timestamp of the spike, and whether TURN was used. With that I can suggest specific next steps and commands to collect diagnostics. Sources: Blog | LiveKit | Voice Agent Architecture: STT, LLM, and TTS Pipelines Explained | LiveKit | A tale of two protocols: comparing WebRTC against HLS for live streaming | ... | Decentraland&#39;s Catalyst: using WebRTC to interact in the metaverse | LiveKi... | Diagnosing Connection Errors with Connection Test Utility, with interactive elements