What's new in Swift Fox, month by month: new device support, new tools, and fixes for the things you've told us about.
Follow the RSS feed · or email [email protected] to get these monthly
The big one this month is RF coverage: Swift Fox now predicts coverage for every sector on your network from LIDAR terrain data, calibrates it against the signal your real customers report, and plots the path to any prospect. Alongside it, a feature to export all your ISP data and another round of netmap beta fixes.
July was mostly about bugfixing and performance improvements to the new network map based on feedback from folks using the beta. Real world use helped us take one 300-site beta network from a 2 GB browser tab down to about 70 MB, and we sped up the map view by moving to MapLibre. The first production Pathbox was deployed, and CGNAT now works properly with more than one edge router.
The rebuilt network map opened to everyone in beta — a year of backend work you can finally see, fast enough that panning around a 5,000-subscriber network feels instant. Alongside it, two tools for working in the field: a real SSH terminal in the browser, and traceroute from anywhere.
Swift Fox got better at figuring out your network on its own: pingboxes passively discover what's cabled to what, the new credentials system tries an ordered chain of credentials until one works, and the core monitoring tasks became always-on daemons.
Two years of MPLS automation work came together this month: Pathbox went from internal project to versioned public releases with a bring-your-own-hardware deb repo, and the rebuilt auto-balancing engine went live in production. The same engine now drives MikroTik cores and Pathbox appliances identically.
Path-MTU problems are some of the most confusing failures to troubleshoot. Swift Fox now measures MTU across your network and works around problems automatically with MSS clamping. Behind the scenes, March was heads-down on the new balancing engine that shipped in April.
The headline this month is pingboxd, a ground-up rewrite of the agent that does the actual pinging, polling, and latency monitoring on your network. Alongside it, the real-time ping test was redesigned to stream results live while you watch.
Mostly behind-the-scenes backend work this month, plus a good number of small fixes for reported bugs.
The network map switched to the new metrics backend for faster loads, vintage pre-M NanoStations joined the supported list, and groundwork landed for the next generation of pingbox.
The paging engine was rebuilt end to end (severities, per-timespan rules, PagerDuty support, rate limits), and backhauls and power devices now have their own metrics and alerting.
Alert thresholds are now editable right on the device pane (per metric, in bulk, and with wait periods), and every device class now reports uptime.
A rebuilt alerting engine with per-device overrides, Linux servers monitored like any other device, Ubiquiti MLO radios, switch bandwidth counters, and smarter CGNAT management.
A hardening month for the new billing stack: tighter card-entry security, automatic tokenization of legacy stored cards, and FCC Form 477 fixes.
The pingbox can now run on a Raspberry Pi you flash yourself, Baicells and generic SNMP devices joined the supported-device list, and Mimosa monitoring got more accurate.
Swift Fox billing moved to Spreedly for processor-independent card tokenization, larger networks can now be organized into regions, and Ubiquiti Wave radios are fully monitored.