TailscaleMetrictailscale.node.latency
Peer latency milliseconds
Google Cloud Monitoring (1)Summary
Network round-trip time (RTT) between this node and a specific peer, typically measured in milliseconds. This gauge tracks connection quality and path performance. Elevated latency indicates network path degradation, routing through distant DERP relays, or congestion. Sudden increases may signal path changes or network issues requiring investigation. Critical for diagnosing performance problems and validating direct peer connectivity.
Interface Metrics (1)
Google Cloud Monitoring
Technical Annotations (79)
Configuration Parameters (9)
ethtool.udp_throughputrecommended: enabledRetriesrecommended: 0Packet Sizerecommended: 57DERP_ADDRrecommended: :3443DERPPortrecommended: 3443OmitDefaultRegionsrecommended: falsederpMap.regions.RegionIDrecommended: null--portrecommended: 41641/etc/defaults/tailscaledrecommended: port configurationError Signatures (3)
relay "code"log patterndirect connection not establishedlog patternvia DERP(log patternCLI Commands (21)
tailscale up --accept-routes --exit-node=my-vps-exit-noderemediationtailscale statusdiagnosticip routediagnostictailscale pingdiagnostictailscale ping <device>diagnosticpingdiagnosticifconfigdiagnostictraceroutediagnostictailscale netcheckdiagnostictailscale versiondiagnosticiperf3 -c serverdiagnosticiperf3 --udp --bitrate 200M -c server -Rdiagnosticmtr serverdiagnosticrsync -ah --progress /radarr-pool/Tenet\ \(2020\)/Tenet\ \(2020\)\ Bluray-2160p.mkv ~/Downloadsdiagnosticscp -C ashayc@<redacted>:/radarr-pool/Tenet\ \(2020\)/Tenet\ \(2020\)\ Bluray-2160p.mkv ~/Downloadsdiagnostictailscale ping 100.64.10.99diagnosticcurl -Iv https://derp.example.com:3443diagnostictailscale ping <node>diagnosticcurl https://controlplane.tailscale.com/derpmap/defaultdiagnosticcurl --silent https://controlplane.tailscale.com/derpmap/default | jq -r '.Regions[] | "\(.RegionID) \(.RegionName)"'diagnostictailscale ping <device-name>diagnosticTechnical References (46)
0.0.0.0/0conceptsubnet routescomponentexit nodecomponenthome DERP serverconceptcoordination servercomponentHard NATconceptDERPcomponenthead-of-line blockingconceptTCP meltdownconceptsubnet routercomponentrelay servercomponentMTRcomponentWiresharkcomponentNICcomponentNAT traversalconcepttailnetconcepttailscaled daemoncomponentdirect connectionconceptaccess control policiesconceptiperf3componentethtoolcomponentDERP relaycomponentJellyfincomponentLinux kernel 6.2componentMTUconceptUDPprotocolWireGuardprotocolport 41641conceptMappingVariesByDestIPcomponentderpMapcomponentTailscale Admin ConsolecomponentOmitDefaultRegionscomponentsymmetric NATconceptCGNATconceptDERP server selectionconcepttailnet policy filefile pathRegionIDconceptDERP servercomponentNAT-PMPprotocolUPnPprotocolQoSconceptpeer relaycomponentNATcomponenthard NATconceptDERP relay serverscomponentPCPprotocolRelated Insights (35)
Packet loss between peers indicates network path instability, congestion, or hardware issues. Even small amounts of packet loss (>1%) can severely impact TCP throughput and latency-sensitive applications, causing retransmissions, increased latency, and degraded user experience.
When a large volume of traffic is routed through a single exit node, it can create a performance bottleneck, saturating the exit node's bandwidth, CPU, or network interfaces. This degrades performance for all users routing through that exit node and creates a single point of failure.
When peer connections are forced to route through DERP relay servers instead of establishing direct peer-to-peer connections, this indicates NAT traversal failures, firewall restrictions, or network configuration issues. DERP relays add latency and consume additional bandwidth, degrading performance and increasing infrastructure costs.
Sustained increases in peer-to-peer latency indicate network path degradation, increased reliance on DERP relays, or intermediate network congestion. This directly impacts application performance for latency-sensitive workloads and user experience.