--- title: Edge monitoring slug: enterprise/edge-monitoring docTags: createdAt: 2026-10-15T14:22:26.062Z --- Several Edges can connect to one Core. Most Edges run on their own server, with its own processing and storage resources. # Check connections and health ![](https://api.archbee.com/api/optimize/JzHznijrWxPLJLDUaV5A1/y4Ox1xGIVsekcbS3a8KTf_overview-camera-health.png "Overview: Portal Physical view and Health alerts") 1. Open **Portal → Administrator → Physical view**. 2. Select the relevant Edge or camera. 3. Review its connection information. 4. Open **Health alerts**. 5. Check alert descriptions and timestamps. ![](https://api.archbee.com/api/optimize/JzHznijrWxPLJLDUaV5A1/nqTUxYl8qXjlIaTn9vVaw_task-camera-health.png "Details: health-alert descriptions and timestamps") Investigate repeated frame or connection problems. Follow [Offline camera or Edge](docId\:uXVDT1CZkhWjKsE9g6vl6) to separate one stream from an Edge-wide fault. If all Edges disconnect at the same time, check the Core certificate's expiration date. [Maintain accounts and certificates](docId\:BxhH1ZaTSVqUYx7mQYT-S) explains how to check it. A Core on the default domain uses addresses under `irisplus.app`, such as `portal.irisplus.app`. For the default certificate, reach out to Irisity for a renewal. # Test the Edge connection to the Core Use these checks when an Edge does not connect, or cannot be added to an account. Run them in an SSH session on the Edge server. Replace `DOMAIN` with your installation domain, by default `irisplus.app`. 1. Check that the Core host names resolve with `getent hosts portal.DOMAIN auth.DOMAIN edge.DOMAIN proxy.DOMAIN registry.DOMAIN`. 2. Check that the Core answers on port 443 with `nc -zv edge.DOMAIN 443`. 3. Read recent device-manager logs with `docker logs --since 30m innovi-device-manager`. Each host name needs a DNS record or an `/etc/hosts` entry that points to the Core. The [Edge installation manual](docId\:mV4fx-LQgpFuodNjsNYi8) lists the required names. The log error `x509: certificate signed by unknown authority` means that the Edge does not trust the Core's certificate. For a custom domain, follow Appendix 1 of the Edge installation manual. Otherwise, reach out to Irisity. # Read Edge resource use Each Edge server reports its CPU and memory use to the Core. The Portal shows these readings; Grafana does not. 1. Open **Portal → Administrator → Physical view**. 2. Select the Edge. 3. Open the **Dashboard** tab. 4. Check **CPU** and **RAM** in the **Device monitoring** chart. The **Dashboard** shows two-minute averages. Keep the average CPU below 80% when the site is busy. The Edge's health state shows **High CPU** at 80% average CPU and **High RAM** at 85% memory use. If CPU or RAM stays high, follow [Reduce Edge load](docId\:nYIItacpKTFMoV4VmhEY2). Short overloads do not appear in two-minute averages. To see them, watch the Edge server with `htop` after you add the full camera load, and while you troubleshoot. The 80% guidance applies to the **Dashboard** average, not to single `htop` readings. Use the [General Edge installation guide](docId\:vzosQe3mNCsVHxj3pjbnH) for local administration access to the Edge server. These readings do not include disk space. See [Keep space for recordings](docId\:nYIItacpKTFMoV4VmhEY2). Edge containers can also run on the Core server, outside Kubernetes. They consume shared resources but do not appear in Core workload charts. # Reduce Edge load When CPU or RAM stays high, lower the work on that Edge. Start with the cameras that need the least detail. Change the frame rate, resolution and codec in each camera's own settings. Change the other options in the camera's settings in the Portal. - **Lower the camera frame rate.** By default, the Edge analyzes at most 8 frames per second. Frames above that rate only add decoding work, so set the camera stream close to 8 frames per second. Do not go below 8, because object tracking then degrades. - **Use a smaller stream.** Point the camera at a 720p or 1080p stream, such as its second stream. The analytics scales every frame down, so extra pixels only cost decoding. - **Prefer H.264.** On most Edge servers, H.265 needs more CPU to decode. The [Edge 10 datasheet](docId\:RY1TPtQUit71woBmcHSPQ) lists fewer H.265 cameras for each stream profile. - **Avoid needless motion in view.** Busy scenes cost more than quiet ones. Aim cameras away from moving trees, water and traffic that no rule needs. - **Turn off analytics a camera does not need.** For example, **Enable High Sensitivity** on a thermal camera makes the Edge also examine small, distant objects. - **Turn off Full frame recording where nobody needs playback.** The Edge then stops writing a recording frame every second for that camera. - **Keep other software off the Edge server.** Other applications compete with the analytics for CPU and memory. Fewer and smaller streams also lower memory use. If memory keeps rising from day to day without new cameras, reach out to Irisity. If the load stays high after these changes, move some cameras to another Edge. In the **Physical view** tree, use the camera's **Detach from device** and **Attach to device** menu items. To plan a larger Edge, reach out to Irisity. # Keep space for recordings Edges store full-frame recordings on their `/storage` drive. By default, the Edge keeps 20% of the drive free by deleting the oldest recordings first. On a drive that has been recording for a while, about 80% use is normal. Use above that means other files are filling the drive. Do not delete files from `/storage` yourself; reach out to Irisity. The Edge deletes only recordings, so other files can keep it from freeing enough space. Recording then stops, and the Portal shows no health alert for this. At each [review](docId\:wqOLXuH2pfp9ig4HyjKG2), check free space on the Edge server with `df -h /storage`. Then play back recent footage from one camera that has full-frame recording, to confirm that recording continues. # Collect Edge logs When you reach out to Irisity about an Edge, include its logs for the incident interval. The Core support bundle does not contain logs from a separate Edge server. 1. Open an SSH session on the Edge server. 2. List the IRIS+ containers with `docker ps -a`. 3. Note the name of each affected container. 4. Run the command below once for each affected container. 5. Transfer the log files through the agreed secure support channel. `docker logs --timestamps --since 2026-10-15T09:10:00Z --until 2026-10-15T09:20:00Z CONTAINER > CONTAINER.log 2>&1` Replace `CONTAINER` with the container name, and the example times with the incident interval in UTC. An Edge runs up to three IRIS+ containers: `innovi-device-manager`, `innovi-analytics-service` and `innovi-digital-io-gateway`. If your account is not in the `docker` group, prefix the commands with `sudo`. Edge logs rotate, and a busy Edge can lose older entries within hours. Collect the logs soon after the incident.