> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/no-events-alerts.md).

# No-Events Alerts

Get an email when a camera, stream, or Workflow stops sending events, so an outage is caught in minutes instead of days.

A No-Events Alert emails your team when a [Use Case](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/use-cases.md) stops receiving events. You choose how long the Use Case can go quiet (ex: 30 minutes) and who to tell. If nothing arrives in that time, they get one email that says when the last event came in.

When a vision system fails, it usually fails quietly. A camera loses power, a stream freezes, a network switch drops, an edge device goes offline after an update, or a Workflow stops running. Nothing raises an error, and the model's accuracy looks the same. Events just stop arriving, and the line keeps running without inspection. Without an alert, the gap often turns up days later, when a shift's numbers come back empty or a defect reaches a customer.

A No-Events Alert is a type of [Automation](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/automations.md), so it lives in the same "Automations" tab and the same activity history.

## Create a No-Events Alert

You need the Manage Vision Events [permission](https://docs.roboflow.com/platform/enterprise-features/role-based-access-control) to create, edit, or delete an alert. Workspace admins have it by default.

<figure><img src="https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-ea48303ba5eb989df7e53a75cc4fa7651395dd5d%2Fno-events-alert-tiles.png?alt=media" alt="The Automations tab of a Use Case with no Automations, showing the Operator feedback preset, No-events alert, and Custom automation tiles"><figcaption><p>A Use Case with no Automations shows three ways to start.</p></figcaption></figure>

1. Click "Vision Events" in the left sidebar, then open a Use Case that is already receiving events. To start sending events, see [Send Events](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/send-events.md).
2. Open the "Automations" tab.
3. If the Use Case has no Automations yet, click "No-events alert". Otherwise, click "New Automation" and set "Trigger" to "No events arrive".
4. Enter a name (ex: "Line 3 camera silent"). The email subject starts with it, so name the alert after what it watches.
5. Set "Alert After" to how long the Use Case can go without events. Pick a number and a unit (Minutes, Hours, or Days), from 15 minutes to 7 days. The default is 30 minutes.
6. Under "Recipients", choose up to 10 people who can act on it (ex: the line's maintenance lead, the shift supervisor, or whoever runs your vision system). Each must be a workspace member with the View Vision Events permission.
7. Make sure "Enabled" is checked, then click "Save Automation".

<figure><img src="https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-ed7c462b47c551e0a8959230b3993c6c673959b0%2Fno-events-alert-editor.png?alt=media" alt="The New Automation editor with Trigger set to No events arrive and Alert After set to 30 minutes" width="375"><figcaption><p>A No-Events Alert needs a name, an "Alert After" period, and recipients.</p></figcaption></figure>

The alert appears in the list with a summary such as "No events received for 30 minutes". A No-Events Alert has no conditions and can only send an email.

## Choose How Long to Wait

Set "Alert After" a little longer than the longest normal gap between events. Too short and you get emails during normal pauses. Too long and the line runs without inspection for longer before anyone knows.

<table data-search="false"><thead><tr><th>How the Use Case sends events</th><th>Suggested "Alert After"</th></tr></thead><tbody><tr><td>Several times a minute (ex: a camera that sends an event for every bottle on a bottling line)</td><td>15 to 30 minutes</td></tr><tr><td>A few times an hour (ex: a part arriving at a station)</td><td>1 to 2 hours</td></tr><tr><td>In scheduled batches (ex: an hourly inspection run)</td><td>A little more than the batch interval (ex: 90 minutes for hourly runs)</td></tr><tr><td>Only during working hours</td><td>Longer than your longest planned downtime (ex: 3 days to cover a weekend), or turn the alert off outside production</td></tr></tbody></table>

Roboflow checks every 5 minutes, so the email arrives between one "Alert After" period and about 5 minutes after that, counted from the last event. With a 30 minute wait and a last event at 9:00, expect the email by about 9:35.

## What Counts as an Event

The alert watches the whole Use Case. Any event counts, from any device, stream, Workflow, or API call, so the alert sends only when the entire Use Case goes quiet. If one of five cameras goes offline while the other four keep sending, no alert sends.

Roboflow uses the time it received each event, not the timestamp the event carries. Events sent late from an [Edge Device Backup](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/send-events.md) count when they arrive. [Operator Feedback](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/operator-feedback.md) and deleted events do not count.

A camera that is knocked out of position or has a dirty lens usually keeps sending events, so it does not trigger a No-Events Alert. To catch image quality problems, see [Automate Camera Quality Monitoring](https://roboflow.com/blog/automate-camera-quality-monitoring).

{% hint style="info" %}
To watch each camera on its own, use [Device Alerts](https://docs.roboflow.com/deployment/self-hosted/enterprise/deployment-manager/setting-up/set-up-device-alerts.md) in Deployment Manager. They watch each device's connection and each stream's camera and inference FPS. A No-Events Alert also covers sources Device Alerts cannot see, such as Workflows in the cloud, your own code calling the API, or a Workflow that runs but no longer sends events.
{% endhint %}

## The Alert Email

Each recipient gets one email per outage. It shows:

* The alert's name in the subject, followed by "No events for" and the "Alert After" period (ex: "No events for 30 minutes").
* The Use Case and when Roboflow noticed the silence.
* "Last event": when the last event arrived, so you can tell whether the outage started minutes ago or yesterday. If no event arrived in the last 30 days, or the last one is older than your Workspace's [retention window](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events.md#data-retention), it says "None in the last 30 days".
* "Alert after": the period you set.
* A "View in Vision Events" button that opens the Use Case.

Times are in UTC, since recipients may be in different timezones.

## When Alerts Send

* One email per outage. An outage starts when a check finds no events for the whole "Alert After" period. You get one email for it, however long it lasts.
* The outage ends when any event arrives. "Recent activity" shows "Events resumed". No recovery email is sent.
* The next outage sends a new email. If events stop again for a full period, that is a new outage.
* A new or edited alert waits first. After you create or save an alert, including a change to the name, the recipients, or the "Enabled" box, it waits one full "Alert After" period before it can send. If the Use Case was already quiet, the email arrives after that period and still shows the real time of the last event.

## Test an Alert

1. Create an alert with a 15 minute "Alert After" period on a Use Case.
2. Stop the device, stream, or script that sends its events.
3. Expect the email 15 to 20 minutes after events stop, and no sooner than 15 minutes after you saved the alert.
4. Start sending again. "Recent activity" shows "Events resumed".

## Track Alerts

Each alert's card on the "Automations" tab shows its "Alert After" period, its recipients, and its daily email cap. If a recipient leaves the Workspace or loses the View Vision Events permission, the card shows a warning and that person is skipped.

<figure><img src="https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-e851f8071f647abf130ad908ca43d8a168c61bac%2Fno-events-alert-card.png?alt=media" alt="A No-Events Alert card showing No events received for 15 minutes, the email action, and a daily email cap of 50"><figcaption></figcaption></figure>

"Recent activity" lists each run. No-Events Alert runs show "No events" in the "Event Type" column.

<figure><img src="https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-a3ca7b0b9035445ef72b6ed9692cb3859225989e%2Fno-events-alert-activity.png?alt=media" alt="Recent activity showing a No events run marked Emailed and another marked Events resumed"><figcaption></figcaption></figure>

<table data-search="false"><thead><tr><th width="200">Status</th><th>Meaning</th></tr></thead><tbody><tr><td>Emailed</td><td>The alert email went out.</td></tr><tr><td>Events resumed</td><td>Events arrived again after an outage that sent an email.</td></tr><tr><td>Cap reached</td><td>The alert used up its daily email cap.</td></tr><tr><td>Usage paused</td><td>Workspace usage is paused or out of credits, so the email is on hold. It sends if usage resumes while the Use Case is still quiet.</td></tr><tr><td>Email failed</td><td>The email could not be sent. See <a href="#limits">Limits</a>.</td></tr></tbody></table>

Runs are kept for 90 days. Creating, editing, and deleting an alert is written to the Workspace [Audit Logs](https://docs.roboflow.com/platform/enterprise-features/audit-logs-and-siem-export).

## Limits

* "Alert After": 15 minutes to 7 days.
* Recipients: up to 10 Workspace members with the View Vision Events permission. Alerts are email only.
* Daily email cap: 50 emails per day by default, up to 1,000. Since each outage sends one email, the cap only matters when a Use Case goes quiet and recovers many times in one day. The day resets at midnight UTC.
* Workspace email limit: each Workspace can send 1,000 emails a day, shared with Workflow email blocks and [Summary Reports](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/summary-reports.md). An alert refused by this limit shows "Email failed" and is retried every 5 minutes while the outage lasts.
* Retries: a failed send is retried on each check. After about an hour of failures, the outage is marked "Email failed" and is not retried.
* Turned-off alerts and archived Use Cases are not checked.
* Each alert email uses 0.01 credits, like other Roboflow email notifications.
* No-Events Alerts are set up in the Roboflow web app. They cannot be created through the API.
