Troubleshooting

Organised by symptom. Two tools solve most problems here: the device status popover (the appliance's vitals — Device health & monitoring) and each camera's Connectivity Diagnosis tab.

Claiming the Edge Processor

The claim fails or the device "can't be found."

Confirm the appliance has power and Ethernet, and has been up for a couple of minutes — the claim requires the device to have recently reported to the cloud.
Double-check the serial number — or use Scan QR and skip typing entirely.
If your network filters outbound traffic, ensure ordinary outbound HTTPS is allowed (network requirements).
Power-cycle the appliance, wait two minutes, retry.

Adding a camera

Test Stream fails.

The Edge Processor couldn't read video at that URL. In order of likelihood:

Credentials — the username/password embedded in the URL. Special characters in passwords often need URL-encoding (@ → %40).
The path — /stream1 vs /Streaming/Channels/101 etc. is brand-specific; look up your camera in the Camera Database or verify in the camera's manual or web admin page.
Reachability — is the camera's IP right, and on a network the appliance can reach? Confirm the camera is streaming at all by opening the URL in a desktop player (e.g. VLC) from the same network.
The camera's own settings — its stream/ONVIF service may be disabled, or it may limit concurrent connections (close other viewers).

"Network Cameras require an AI Edge Processor to be online."

Cameras attach to a processor, so the processor must be online first — check its status, then retry.

Streams and live view

A camera tile says the camera is offline.

The appliance can't currently read the camera's stream. It retries automatically with backoff — so if the camera just rebooted, it will come back on its own. If it stays offline:

Open the camera's Connectivity Diagnosis tab — it tests the path from the appliance to the camera and usually names the problem.
Check the obvious physical causes: camera power (PoE port, injector), network cable, a changed IP address (which breaks the stream URL — this is why static IPs/reservations are recommended).
If the camera responds but the stream doesn't, power-cycle the camera and use Test Stream to verify.

Live view is choppy or stuck on low quality.

Open the device status popover and run the client speed test — it separates your connection from the facility's. The module tells you which side is constrained when it detects a slow path, along with recommended speeds for the camera count.
Reduce the number of simultaneous tiles, or let Auto quality do its job rather than forcing High/Native.
Remote viewing falls back to a buffered mode when a direct path can't be established — that adds a few seconds of latency by design. On-network viewing should always be near-instant; if it isn't, check the appliance's uplink.
A "Connected via Wifi" badge means the appliance itself is on Wi‑Fi — move it to Ethernet.

Streams pause on their own.

That's the inactivity guard ("Livestream is currently paused due to inactivity."). Click Resume Livestream, and for dedicated monitoring screens enable Never ask to continue watching in Settings → Live View (Live view).

Recordings

Motion happened but there's no clip.

Is Recording enabled for the camera, and in the right mode? The Dashboard activity strip shows "Recording Disabled" badges.
Check the motion zone — motion outside the zone doesn't trigger. Open the editor and confirm the zone covers the area (Motion zones).
Check the clip isn't just still uploading — cards show "Uploading..." until the transfer lands.
If the camera was offline at the time, there is nothing to record — correlate with the activity strip.

A clip I need is gone.

Recordings age out per your retention policy (default one month). Expired footage is not recoverable. Two mitigations: retain incident footage immediately, and check Timelapse — its retention is much longer and may still show the scene at one-minute resolution.

Uploads

"ITEMS QUEUED" is high and not falling.

A backlog forms during heavy motion, after an internet outage, or while someone watches live streams (uploads pause for live view by design and resume a few minutes after). A queue that never drains means the facility's upload bandwidth doesn't match what the cameras produce: check the facility's measured upload speed in the header, then reduce production (motion mode instead of continuous for busy cameras, tighter motion zones) or increase bandwidth.

Storage Used on the appliance keeps climbing.

Same root cause — clips are being produced faster than they upload. The appliance protects itself (oldest already-uploaded files are cleared first), but fix the bandwidth mismatch before a long outage stacks the odds.

The appliance

Temperature is high / "Device Overheating" appears.

The appliance throttles at 77 °C. Give it airflow: clear the vents, get it out of sealed cabinets or direct sun, and away from other hot gear (placement). Throttling degrades AI processing first, then streaming.

The whole device is offline.

Power and internet, in that order. The health beacon is designed to survive software updates, so a genuinely offline device isn't "just updating." If footage matters right now, remember: an appliance with power but no internet is still recording locally and will upload the backlog when connectivity returns.

"Provisioning device changes" / "Device updates are currently in progress" won't clear.

Configuration pushes settle within minutes; nightly updates within the maintenance window. If a device is stuck for hours, use Restart Device, then contact support.

When to contact support

Include the facility name, the device's serial number, camera names, and timestamps — and leave the device powered on so its telemetry and logs are reachable remotely. Support can inspect the appliance's health history without a site visit.