Supported cameras and recorders can push events to Hermetry as they happen: motion, tampering, video loss, disk faults, illegal logins and network faults. Each event becomes a reading of the camera’s device, so alarm rules, flows and charts use it like any other value. Events are off by default and switched on per endpoint.
Events are available for the HTTP protocol of supported cameras and recorders. RTSP cameras have no event stream.
Switch events on
- Open the camera’s device page and its Connection tab (or the Camera tab), and select Edit endpoint / Edit connection.
- Switch on Receive device events (“Keeps one long-lived connection to the device for its events (motion, tamper, video loss, disk faults). Each event counts as a message of the plan.”).
- Save. The connection test also opens the event stream once and warns if it cannot:
| Warning | Fix |
|---|---|
| “The account may not open the event stream. Give it the right to receive alarms, or switch events off.” | Give the camera user the right to receive alarms or events on the camera. |
| “This device has no event stream, so device events cannot be received from it. Switch events off for this endpoint.” | The device does not support it. |
| “The event stream could not be opened on this device. It is retried in the background.” | A temporary problem. |
The camera must also be set up to raise those events (for example, motion detection must be on for the area you care about). That is done on the camera.
For a recorder, switch it on once on the recorder’s endpoint: it serves all its channels.
Event keys
Each event is stored as telemetry on the device it belongs to, under a key event.<kind>: 1 while the condition is on, 0 when it ended.
| Key | Event | Label in the ready profile |
|---|---|---|
event.motion | Motion detection | Motion |
event.tamper | Lens or housing covered, moved or tampered with | Tamper |
event.video_loss | Video signal lost | Video loss |
event.disk_error | Disk error, unformatted disk, failed array | Disk error (event) |
event.disk_full | Disk full | Disk full |
event.login_failure | Login with a wrong user name or password | Illegal login |
event.network | IP address conflict, network card fault | Network fault |
event.other | Any other event; the device’s own word is kept in the details | — |
Which device gets it: a camera’s events go to the camera. A recorder’s event for a channel goes to that channel’s device; an event without a channel (disk, login, network) goes to the recorder’s own device (channel 0). Events of a channel that is not added in Hermetry are dropped.
Time: the event’s own time from the device, corrected by the clock drift Hermetry measures. If the device’s clock is more than 5 minutes off after that, or unreadable, the arrival time is used.
Bursts are folded. Within 10 seconds, the first event of a kind for a device is delivered at once; repeats are folded, and the last one is delivered at the end of the window only if it says something new (for example the end of the condition). A same-state repeat within a minute is a duplicate. One folded message counts once.
Ready alarm rules
The ready profile for supported cameras and recorders includes alarm rules on these keys:
| Rule | Severity | Raised when | On by default |
|---|---|---|---|
event_tamper | Major | event.tamper == 1 | Yes |
event_video_loss | Major | event.video_loss == 1 | Yes |
event_disk_error | Major | event.disk_error == 1 | Yes |
event_disk_full | Warning | event.disk_full == 1 | Yes |
event_login_failure | Warning | event.login_failure == 1 | Yes |
event_network | Warning | event.network == 1 | Yes |
event_motion | Info | event.motion == 1 | No: motion is frequent; switch it on where it matters |
Major alarms notify the Admin role. Each alarm clears when the device reports the end of the condition. A login failure has no end event: acknowledge or resolve it by hand.
The same profile has three rules on its polled health readings: disk_error (Major, a disk or card is not ok), device_reboot (Warning, the device restarted since the last poll or two) and device_offline (Major, three polls in a row failed).
Change or add rules on the profile; see Alarm rules on device profiles.
Connection state of the event stream
The device’s Connection tab shows Device events with a state: Connected, Reconnecting, Failed, Waiting or Off, the Last event and, after a failure, Events error and Next attempt. The endpoint list shows the state in its Events column.
If the stream drops, Hermetry reconnects with growing pauses (2 seconds, doubling, at most 5 minutes). A failure that cannot fix itself, such as a wrong password, waits at least 15 minutes between attempts so the camera does not lock its account. When you change the endpoint’s settings or password, the stream restarts at once. A stream that stays silent for 45 seconds, without even a heartbeat, is treated as dead: “The device stopped sending events: not even its heartbeat arrived.”
Quota
Each delivered event message counts as one message of your plan. A burst folded into one message counts once. When your organisation is over its message limit, events are dropped. The number of event streams your organisation can hold open at once is limited by the platform.