Notifications

In the sidebar, go to Settings > Notifications. This decides which built-in system events this installation reports, where they go, and how often — separate from the user-defined Alerts.

Delivery is over ntfy.sh (or a compatible self-hosted ntfy server) — a push-notification service, not e-mail or Telegram. Each channel is one ntfy topic; the built-in watchdog and the System status ("doctor") page both read this configuration directly to know what to report.

Channels

A channel is one ntfy destination. This test system already had one configured:

Notifications: channels use ntfy.sh; the topic is masked in the UI itself once saved.
FieldMeaning
NameA label for this channel, shown on its button above the form.
ntfy serverThe ntfy server to publish to — https://ntfy.sh by default, or a self-hosted one.
TopicThe ntfy topic name. Left empty, one is generated randomly. Once saved, the product itself only ever shows it masked — see the warning below.
Access tokenAn optional ntfy access token for a protected topic; shown as Not set or Set (hidden), never in the clear.
Send test messagePublishes one real test notification to this channel, immediately.
Warning

On the public ntfy.sh server, a topic is only as private as its name: anyone who knows (or guesses) it can read every notification sent to it. Use a long, random topic name, or an access token on a self-hosted server, for anything sensitive.

Events

Below the channel form, one row per built-in event — each with its own internal code (e.g. ast_down, pg_down) shown next to the label, and its own priority, repeat interval, daily cap, quiet hours and an "all-clear" option once the condition goes away.

Events: the built-in catalogue — infrastructure, traffic-shape and resource events, each independently tunable.
ColumnMeaning
PriorityMinimal, Low, Default, High or Urgent — an ntfy priority level, which affects how the notification is presented on the receiving device.
Repeat (min)Minimum minutes between repeats of the same still-ongoing event.
Max per dayCaps how many times this event can notify in one day.
Quiet hoursAn hour range (0–23) during which this event stays silent.
Send all-clearAlso send a notification once the condition is resolved, not just when it starts.

The note on the page explains a key wrinkle: "Rows you never touch use the built-in defaults. A key ending in * covers a whole family, for example every CID pool." — you only need to add a row here for an event you want to tune away from its default.

Call ceiling for the watchdog

A single number, Concurrent calls, used only when the licence itself sets no channel limit — the watchdog then warns at 80% and reports critical at 95% of this ceiling. Left at 0, that particular alarm stays silent, since there is nothing to measure against.

Delivery log

What was actually delivered — not what was attempted. A failed delivery does not count against an event's daily cap.

Delivery log: what was sent, and — as here — what was suppressed because the daily limit was already reached.
StatusMeaning
DeliveredSuccessfully sent to ntfy.
FailedThe send itself failed (see the Details column for the HTTP code/error).
Daily limit reachedSuppressed — this event's Max per day was already used up.
Quiet hoursSuppressed by that event's quiet-hours window.
Switched offSuppressed because the event, or the whole channel, is disabled.

Check

Use Send test message after changing a channel's server, topic or token to confirm delivery still works; after tuning an event, check the Delivery log the next time it should have fired to see whether it was delivered or suppressed, and why.