Fleet & Remote Tools
Link servers over SSH and run diagnostics on any node or ad-hoc IP, using a saved sudo password through the helper.
Unattended recovery
See resource and temperature history together
An enrolled server's Behaviour & History view keeps CPU, memory, disk and available temperature readings at five-minute resolution for up to 30 days. Temperature has its own Celsius scale so it remains readable beside percentage-based resource use. Reboot and watchdog markers stay aligned across the timeline, making it easier to connect a heat or load change with a recovery event.
Escalate only after service recovery is exhausted
Open a pinned server and expand Service Watchdog. The helper on that server should remain the first line of recovery: it checks the affected service and uses its bounded restart budget locally. FrontierStack's optional recovery ladder starts only after a new Recovery exhausted event. Old events and watchdog actions suppressed by the helper's own anti-loop rules are never replayed as a new reboot.
One reboot, then a guarded power cycle
Enable Reboot from FrontierStack to let the Mac issue one verified host reboot after the delay you choose. For the final step, link a controllable outlet in Power Control, select it in the server's watchdog panel, and set the full-outage timer. FrontierStack waits at least five minutes and cuts power only while the Mac's network is healthy and the server has no SSH, helper monitor, ping, ARP or watched service-port response. One incident can produce only one hard cycle.
Keep the final watch running without the Mac
A wedged server may be unable to control its own outlet. Choose a different, always-on enrolled server as the Recovery observer. Its updated FrontierStack helper probes the target's private address by ping and several TCP ports, so the final power watch continues while the Mac app is closed. Before cutting power it also requires a successful status response from the private power controller, saves an anti-loop receipt, and then waits until it has seen the target recover before another incident can be considered.
Treat power recovery as an emergency control. The observer receives no shell access and no credential for the watched server. It receives only the private target address, probe ports, timer and selected power-controller settings. A Home Assistant setup necessarily includes the selected entity and token; the helper stores this limited policy in protected root/system state. Target and controller addresses must be literal private, link-local or VPN IP addresses. Generic HTTP controllers need a separate status URL. Test the complete path during a maintenance window.
Reboot a server that is on but locked up
In the same Service Watchdog panel, Lockup reboot (monitor 1.23.0 or later) lets the server reboot itself when SSH stops answering; the monitor waits for sshd's greeting, not just an open port, because a wedged Mac often accepts the connection and then drops it; or when it loses the network and can reach neither its default gateway nor any extra target you list. Choose how many minutes of continuous failure count (10 by default). The server reboots at most once per interval, only after the check has worked once since the monitor started, and only with Allow actions and Allow reboot on. The panel shows each check's live state, the gateway in use and the last automatic reboot.
From monitor 1.24.0 a reboot sentinel runs beside the monitor on macOS and Linux: the same binary as root, with no network listener and no token. It follows the monitor's own policy and permissions and acts three minutes after the monitor's reboot should have happened, calling the kernel's reboot directly, so it still works when the server can't start new processes. The monitor also samples process counts every minute against the kernel and per-user limits, with a 24-hour chart and a warning near the limit.
Plan for the side effects. If the gateway itself goes down, every server with the network check on reboots, at most once per interval. A lockup reboot skips a graceful shutdown, so databases run crash recovery on the next start. A kernel hang stops the monitor and sentinel too: on Macs also turn on Restart if it freezes in Headless Mac Setup, and keep a smart plug or recovery observer for power-level recovery.
See a silent server as it was last seen
When a linked server stops answering, its pane shows what FrontierStack last saw while it answered; OS, hardware, addresses including stable IPv6 and Tailscale, DNS, connection, the running services, any OS-scheduled reboot and the monitor's settings; dimmed with a clock icon under an orange Not answering line. New IPv6 and Scheduled reboot rows are live while the server answers. The record stays on your Mac, never includes tokens or alert URLs, and is forgotten when you remove the server.
This is the web edition of FrontierStack's in-app Help. Open the same topic inside the app for live, clickable controls.
Run it from your Mac.
FrontierStack installs, monitors and secures services on this Mac and on linked servers; and keeps your passwords and keys out of any AI's sight.
Download FrontierStackApple notarized · Safe & secure · macOS 13 Ventura+