See the heartbeat of your network.

Dropped packets now line up correctly. A lost ping was timestamped when its timeout expired rather than when the probe was sent, which placed its red bar a second late — on top of the next successful reply instead of in the gap where the loss happened. Samples are now stamped at send time.

History keeps itself tidy. Old samples were only cleared at launch, so an app left running in the menu bar for weeks never pruned, and the database file never shrank even after deleting rows. Pruning now runs on a timer and reclaims the space.

Saved state survives. A target left switched off for over a week lost the snapshot that lets it redraw instantly when re-enabled. Those snapshots are no longer aged out.

Traceroute accuracy. A run cut short by a network error was treated as a finished path, discarding real hops and recording an intermediate router's response time as the destination's. It is now recorded as a single lost measurement.

Smaller fixes. Switching between targets no longer carries the previous one's scroll position across; live values keep updating while the window is being resized; clicking the already-selected target in the sidebar returns to its detail view.

First public release.

New — Service Check targets. Plenty of hosts firewall ICMP, so a ping says they're down when the service behind them is fine. A Service Check measures what actually matters: a TCP handshake to host:port, or an HTTP(S) request timed to response headers. It reuses the same charting, history, and alerting as every other target type.

New — Overview dashboard. Every active target on one screen, sized to its type, with a sweep heatmap for range scans.

New — Jitter and connection quality. Mean inter-packet delay variation plus an ITU-T G.107 E-model MOS estimate, shown alongside the latency stats.

Rebuilt probing engine. PingGrapher no longer shells out to ping and traceroute. All probing now happens in-process over unprivileged ICMP sockets, which means no helper processes, no output parsing, and true per-packet timing. The app is also fully sandboxed.

Upgrading from an earlier build: your targets and preferences carry over automatically. Historical latency graphs do not — sandboxing moves the app's storage into a container, and previously recorded history stays outside it. New history starts recording immediately.

Note for anyone who downloaded 1.0 on launch day: the app's identifier changed to com.sixdoes.PingGrapher shortly after release. macOS keys an app's storage to that identifier, so the current build starts with an empty history and target list rather than picking up the earlier one's. Re-adding your targets is the only step; the old data can be removed with rm -rf ~/Library/Containers/net.pickering.PingGrapher.

Also on the Mac App Store. PingGrapher is available both as a direct download here and from the Mac App Store, always at the same version. Identical builds from the same source — pick whichever you prefer.

Requires macOS 14 Sonoma or later. Universal — native on Apple silicon and Intel.