tailscale.peers.online

Online reachable peers

Dimensions:None

Available on:

Prometheus (1)

Summary

Current number of peer devices that are actively connected and reachable from this node's perspective. This gauge reflects immediate connectivity state. Low counts relative to total peers indicate widespread connectivity problems or network partitioning. Sudden drops signal connectivity failures requiring investigation. Essential for validating peer-to-peer connectivity and detecting network segmentation.

Interface Metrics (1)

Prometheus

tailscale_online_peers

Number of peers currently online and reachable

Dimensions:None

Sources

tailscale_online_peersgithub.com

Technical Annotations (14)

Configuration Parameters (2)

Retriesrecommended: 0

Zero retries for immediate alerting on connectivity loss

Packet Sizerecommended: 57

ICMP packet size for ping monitor

Error Signatures (2)

ssh: connect to host 100.x.x.x port 22: Operation timed outlog pattern

UNREACHABLE!log pattern

CLI Commands (3)

tailscale statusdiagnostic

tailscale status --json | jq '.BackendState, .Self.DNSName, .Self.Online'diagnostic

tailscale debug peer-relay-sessionsdiagnostic

Technical References (7)

tailnet policy filecomponentACLscomponenttailnetconceptdevice approvalconcept100.64.0.0/10conceptCGNATconceptpeer relaycomponent

Related Insights (9)

Inter-device communication failure within tailnetcritical

▸

ACL policy blocks inter-device connectivity in tailnetcritical

▸

Ping monitor failure indicates network connectivity losscritical

▸

Node missing from tailscale status indicates authorization or tailnet mismatchcritical

▸

Excessive concurrent connections cause tunnel failures requiring restartcritical

▸

Peer relay service running but not accepting sessionswarning

▸

High peer connection count approaching platform limitswarning

As the number of active peer connections approaches Tailscale's per-node connection limits (typically 1000+ concurrent peers), performance degrades due to increased connection state overhead, CPU utilization, and memory consumption. This can lead to connection establishment failures, increased latency, and daemon instability.

▸

Low peer online ratio indicating tailnet health issueswarning

When the ratio of online peers to total peers drops below expected levels, this indicates a significant portion of the tailnet is unreachable. This could be caused by widespread connectivity issues, mass daemon failures, or infrastructure problems affecting multiple nodes.

▸

Stale node presence indicating connectivity or coordination plane issueswarning

When tailscale.node.last_seen timestamp becomes stale (not updating for extended periods), it indicates the node is not successfully communicating with the Tailscale coordination server or peers. This can be caused by network partitions, firewall rules blocking control plane traffic, or coordination server issues.

▸