# Troubleshooting Source: https://cumulocity.com/docs/2026/edge/troubleshooting-edge/ Sector: Edge Release: 2026 Description: Resolve common Edge problems, such as installation or update loops, pending service IPs, pods stuck in a pending state, and Edge missing from the cloud tenant. ## Edge CR status InstallLoopBackOff or UpdateLoopBackOff After deploying or updating Edge, if the Edge CR status is `InstallationLoopBackOff` or `UpdateLoopBackOff`, as part of the output of `kubectl get edge c8yedge -n c8yedge`: ```shell NAME DOMAIN NAME VERSION STATUS c8yedge myown.iot.com 2026.0.0 InstallLoopBackOff ``` Describe the Edge CR (cumulocity-iot-edge) with the command `kubectl describe edge c8yedge -n c8yedge`. In the events section, you will find the reasons for the install or update failure. For example: ```yaml Name: c8yedge Namespace: c8yedge API Version: edge.cumulocity.com/v1 Kind: CumulocityIoTEdge Status: Deploying Generation: x State: InstallLoopBackOff Events: Type Reason Age From Message ------ ------ ---- ---- ------- Normal Installing 8m34s cumulocityiotedge installing Cumulocity Edge version tagged as 2026.0.0 …… …… Normal InstallLoopBackOff 1m00s cumulocityiotedge waiting for mongo server to be ready ``` MongoDB requires a CPU that supports AVX instructions. Ensure that the CPU type of the Kubernetes node supports AVX instructions. Use the command `lscpu` to check whether the CPU supports AVX instructions. Check the logs of the Edge operator, MongoDB and core pods to get more insights. The Edge operator logs: ```shell kubectl logs -n c8yedge deployment.apps/c8yedge-operator-controller-manager -c manager ``` MongoDB logs: ```shell kubectl logs -n c8yedge --all-containers --prefix statefulset.apps/edge-db-rs0 ``` Core logs: ```shell kubectl logs -n c8yedge --all-containers --prefix statefulset.apps/c8ycore-sts ``` ## Pending external cumulocity-ontoplb service IP The external IP is not assigned to the `cumulocity-ontoplb` service. Output of `kubectl get service cumulocity-ontoplb -n c8yedge`: ```text NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE cumulocity-ontoplb LoadBalancer X.X.X.X **REDACTED 443:32443/TCP,8443:32442/TCP,1883:32083/TCP,8883:32084/TCP ... 12m ``` For more information, see [Assigning an external IP](https://cumulocity.com/docs/2026/edge/manage-edge/#external-ip). ## Edge not accessible and pods in pending state If you installed Edge by using the `c8yedge` tool, the instance is not reachable, and `kubectl get pods -A` shows output similar to the following, the node might be under disk pressure. ```shell NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES apama-ctrl-scope-edge-deployment-7f4796c6c5-shnrr 0/1 Pending 0 2d21h c8ycore-sts-0 0/2 Pending 0 2d21h ``` When you describe a pending pod, you see a scheduling error similar to the following: ```shell kubectl describe pod -n # ... Warning FailedScheduling 12m (x832 over 2d21h) default-scheduler 0/1 nodes are available: 1 node(s) had untolerated taint {node.kubernetes.io/disk-pressure: }. preemption: 0/1 nodes are available: 1 Preemption is not helpful for scheduling. ``` When installed with the c8yedge tool, the Cumulocity Edge runtime uses **K3s** with an embedded kubelet. The kubelet maintains disk availability through Kubernetes **Node-pressure eviction**, using `/var/lib/rancher` as the primary node filesystem path. When free disk space drops below the configured eviction thresholds, the node reports **DiskPressure**. In this state, the kubelet proactively evicts running pods and blocks scheduling of new pods to preserve node stability. As a result, workloads can appear in `Pending`, `Evicted`, or `ContainerStatusUnknown` states. To resolve this issue, reclaim sufficient disk space on the affected node. After **DiskPressure** clears, normal scheduling resumes and workloads recover automatically. To inspect node conditions and capacity, run the following command: ```shell kubectl describe node ``` ## Edge not showing up as a device in the cloud tenant Edge is unable to register itself as a device in your cloud tenant. Describe the Edge CR `kubectl describe edge c8yedge -n c8yedge` In the events section, you will find an event with a reason. For example: ```yaml Name: c8yedge Namespace: c8yedge API Version: edge.cumulocity.com/v1 Kind: CumulocityIoTEdge Status: Deployed Generation:  x State: Ready Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal Installing 8m34s cumulocityiotedge installing Cumulocity Edge version tagged as 2026.0.0 …… …… Normal Ready 1m00s cumulocityiotedge Cumulocity Edge installation is complete, and it's now running version 2026.0.0 Warning FailedToRegisterCloudDevice 0m30s cumulocityiotedge Device failed to connect to , you might not have uploaded the device certificate to Cumulocity ``` Possible reasons: * You didn’t add the Certificate Authority (CA) certificate to the trusted certificate list of your tenant. For completing the registration process, sign into your cloud tenant and follow the steps described in [Managing trusted certificates](https://cumulocity.com/docs/2026/device-certificate-authentication/managing-trusted-certificates/) to add the Certificate Authority (CA) certificate to the trusted certificate list of your tenant. For more details, see [Device certificates](https://cumulocity.com/docs/2026/device-certificate-authentication/device-certificates/). For more information see [Connecting Edge to the cloud](https://cumulocity.com/docs/2026/edge/connecting-edge-to-cloud/). * The specified cloud tenant domain is not reachable or incorrect. ## Contacting product support If you must contact [product support](https://cumulocity.com/docs/2026/additional-resources/contacting-support/), include the output of the diagnostics dump. For more information about accessing diagnostic logs, see [Accessing logs](https://cumulocity.com/docs/2026/edge/edge-operations/#accessing-logs).