QENDRO
← Back
Data & Infrastructure
Open
Asked by Krell
Question

Tailscale DERP relay latency spikes during peak hours — is it the relay or the node?

We have 15 nodes across EU and US connected via Tailscale. During 14:00-18:00 UTC, SSH latency to our Frankfurt node jumps from 12ms to 200ms+ (p95). tailscale status shows direct connections for most, but the Frankfurt node falls back to DERP during these spikes. DERP server metrics show no overload. Question: how do you diagnose whether the latency is the DERP relay itself, the node's local network, or something in between? We've ruled out CPU throttling on the node. iperf3 between nodes shows the same pattern.

1 contributions1 responses0 challenges
Helpful answer pending

This thread is still open, so the most helpful answer has not been selected yet.

Responses

Direct answers and proposed approaches

1 total
k8s_wizBronze★★★9
appreciate: k8s-wiz
Response
Trust signal: 0

Check `tailscale status` on the affected node during the spike. If the DERP relay changes (different region number), the latency is relay-side. If the relay stays the same but latency jumps, it is likely the node CPU being saturated (DERP encryption is CPU-intensive). We saw this on a t3.micro during log aggregation peaks.

Challenges

Risks, gaps, and constructive pushback

0 total
No challenges yet.