Set up your plant
Email alerts
What Started watches your machines and emails your owners and admins when one needs attention. There are four kinds of alert. Each machine has at most one open alert of each kind; open alerts are listed above the machine table on the home page of the app.
The four alerts
| Alert | Opens when | Closes when |
|---|---|---|
| No signal | A machine that has sent pings misses two in a row, never sooner than 2 min | The next ping arrives |
| Down | The machine is set to down, by an operator at the QR sticker or by a ping | It is set to running, idle, waiting or off |
| Not producing | The machine is running in a shift but its cycle count has stopped rising | The cycle count moves again, or anyone logs a different status |
| Out of range | A metric value arrives outside its normal range | A value inside the range arrives |
No signal
Machines are expected to ping every minute, so a machine goes to No signal about two minutes after its last ping. The board shows how long it has been quiet. When pings come back, the machine returns to the status it had before.
Machines that have never sent a ping are not watched, so a plant that only uses QR stickers gets no no-signal alerts.
Down
Any reason that puts a machine in down opens the alert, as does a ping with "status": "down". Which reasons set down is up to you under Settings → Reasons. Waiting and idle reasons do not open an alert.
Not producing
For machines that report a cycle count. Open the machine and set Minutes without a cycle under Not producing. When the machine has been in run for that long during one of its shifts with no new cycle, What Started sets it to the plant's no-cycles reason ("No cycles (unexplained)" unless you changed it on the Reasons page) and emails you. Leave the field empty to turn it off.
- 2 to 1440 minutes. The page suggests the time for three parts at the machine's target or ideal rate.
- It never waits less than two pings, because a counter is only seen as often as the machine pings.
- Outside the machine's shifts it does not fire. A machine with no schedule is checked around the clock. See Shift schedules.
- Machines counted by hand have no counter to watch, so the check is not offered for them.
- Needs a plan with machine counts. See Counts, scrap and rates.
Out of range
For metrics you send with pings, such as a temperature. Turn on Alert for a metric and set its normal low and high under Settings → Metrics. You get at most one email per machine and metric every 30 minutes. Needs a plan with machine metrics. See Live values, charts and range alerts.
How often checks run
No signal and not producing are checked every minute. Down and out of range open the moment the status or value arrives. Emails go out once a minute.
An alert that closes before the next email run is not emailed, so a stop of a few seconds does not wake anyone.
Who gets the emails
Every owner and admin of the plant, at the email address they sign in with. Members and operators do not get alert emails. There is no per-person setting yet: to stop someone getting them, change their role under Settings → Team. See Team and roles.
Turning alerts off
- Down and No signal are on for every machine and cannot be turned off. An archived machine is no longer watched.
- Not producing: clear Minutes without a cycle on the machine.
- Out of range: turn off Alert for the metric. That also closes its open alerts.
What the email looks like
Plain text, one alert per email. The subject starts with your plant's name:
[Acme Plant] Press 4 is down, with the reason if one was logged and the time it went down.[Acme Plant] No signal from Press 4, with the time of the last signal.[Acme Plant] Press 4 is not producing, with the time of the last cycle and the reason What Started set.[Acme Plant] Oven temp out of range on Press 4, with the value, the normal range and when it started.
Each email has a link to the machine's page. Times are in UTC. The machine's area is shown after its name.
Your own systems
The same alerts go out as alert.opened and alert.resolved webhooks, with kind no_signal, down, not_producing or metric_range, and can be read from GET /api/v1/alerts in the read API. Use these to send alerts to a phone, a chat channel or an andon system.