Alert statuses and thresholds¶

Alerts are about the dashboard's own behaviour: when to make a noise, when to change the LED colour, and when to scroll the list to a service that just got worse. They never touch a provider account.

What raises an alert¶

A percentage is mapped onto the status vocabulary, so a provider that reports a percentage needs no code of its own:

Threshold	Status	LED
`warning_percent` (default 80)	`near_limit`	yellow
`critical_percent` (default 95)	`critical`	red
100%	`rate_limited`	red

An account takes the status of its worst window. Two thresholds apply to every service by default, and each service can override them.

Only a fresh, available window alerts:

a stale window keeps its stale presentation and raises nothing,
an unavailable window never alerts,
a window whose reset has passed never alerts,
a window that resets within the next 30 minutes never alerts: it is about to clear.

The sound¶

The sound is off until you turn it on. A desk device must not beep unannounced, so the Sound on a new warning box in the Alerts section of the admin page is unticked until you tick it. The LED, the thresholds and the pop to top work with the sound off.

The device tracks every service separately, so a second service getting worse still speaks while another one is already critical.

Event	Sound
A service becomes a warning	one chirp
A service becomes critical, rate limited or offline	two chirps
A warning that persists	nothing, it is not repeated
A service recovers	nothing, recovery is silent
The first reading after boot	nothing, it is a silent baseline

A long press on the Presto mutes the sound. The mute ends by itself, and the screen says how long is left - see the mute that expires.

The mute that expires¶

A long press on the Presto mutes the sound, so a desk device cannot chirp in the middle of a meeting. The mute is not forgotten: it ends by itself.

Setting	Default	What it does
`mute_hours`	`4`	How long the mute lasts, in hours. 0 to 48. 0 means it lasts until you press again

While the sound is muted, a small chip says how long is left:

The chip	What it means
`MUTED 3H`	The mute ends in about three hours
`MUTED 45M`	The mute ends in about three quarters of an hour
`MUTED`	The mute has no end: it lasts until you press again

The chip is in the bottom-right corner of the Home screen and of every card, and it disappears when the mute ends.

A long press while the sound is muted unmutes it at once.
The mute ends silently. The statuses were tracked while it was on, so the Presto does not chirp for an alarm that began during the mute.
A restart clears the mute: it is not stored anywhere.
Changing mute_hours while a mute is running applies to the next mute, not to the running one.

The Presto has no clock of its own, so it counts the time with its own tick counter. The daemon only sends the length.

The LED¶

The LED shows the worst service on the dashboard, not only the account on the open card:

Colour	Meaning
red	critical, rate limited or offline
yellow	a warning
blue	a service in use
green	everything else

A service that is merely in use never hides a warning.

Quiet hours¶

Quiet hours are off by default. The daemon judges them with its own local time and sends the flag in the payload, so the device needs no clock and no time zone. While quiet hours are on, the device is silent and its light is dimmed - the colour is kept, so a warning is still visible.

Setting	Default
`quiet_hours`	off
`quiet_start`	`22:00`
`quiet_end`	`07:00`

The times are the local time of the machine the daemon runs on.

Pop to top¶

When a service gets worse, the Home list can scroll so that its row is on screen. The rules exist so that the list never jumps while you are using it:

the alarm must hold for 30 seconds before the list moves,
a service moves the list at most once in ten minutes,
the list never moves within 20 seconds of a touch or a scroll,
a new alarm only acts when it is worse than the worst one already shown,
your own position comes back a minute later, unless you scrolled since.

Pop to top is on by default and can be switched off in the Alerts section of the admin page.

The master switch and per-service settings¶

Alerts in the Settings tab has one master switch that turns alerts off for every service at once. While the master is off, every per-service control is disabled but keeps its value, and the service cards say alerts off (global).

Each service can then override the master:

alerts on or off for that service, and
its own warning and critical percent. A blank field means use the global value; the field shows the global value as its placeholder.

These per-service settings are stored in their own rows, keyed by the service ID. Switching the alerts of one service never rebuilds the source and never restarts its readers, and they survive an edit, a rename and a rebuild. They are removed when the service is deleted.

A service's own thresholds are used only when the pair is usable: the warning percent must sit below the critical one. When it does not - for example because the global value changed underneath it - the service falls back to the global pair, the daemon logs one line, and the admin page says override ignored.

What "alerts off" means¶

For that service, or for all of them when the master is off:

no chirp,
no LED escalation,
no pop to top,
no part in the overall accent.

It never hides data. The row, the numbers, the true threshold status colour and the service's own card are unchanged. The daemon still computes the service's true status.

The Presto can switch one service's alerts¶

The bell in the corner of a service card switches that one service's alerts on or off. It is the only thing the device ever writes, and it writes to your daemon, not to a provider. See Alerts and the card bell.