Troubleshooting
Make sure that the IP addresses and the virtual machine (VM) names in the configuration file are not currently used, when deploying a new deployer or a new Cisco Smart PHY cluster.
For deployers, the VM name is the same as the deployer name.
For single-node clusters, the VM name is the cluster-name with -ops
.
For a multinode cluster, there are 12 VMs. Their names are cluster names with -master-n
, -etcd-n
, -infra-n
, and -ops-n
, where n
is 1, 2, and 3.
Troubleshoot Deploying a New Deployer
-
Check if the VM is created on a vCenter.
-
Log into the deployer VM using SSH with the correct username and public key file. ssh -i <private-key-file> <deployer-user>@<deployer-address>
-
Use kubectl command to find the internal IP address of the Operation Center service:
kubectl get svc ops-center-smi-cluster-deployer -n smi
-
Look for the CLUSTER-IP field in the output. Use it to SSH into the deployer: ssh admin@<cluster-ip> -p 2024
-
Check whether the product tar files available in the offline-products
directory are downloaded to the deployer:software-package list
Troubleshoot Deploying a New Cisco Smart PHY Cluster
-
Check if the configuration for Cisco Smart PHY clusters is pushed to the deployer: show running-config
-
Monitor the deployment status from the deployer: monitor sync-logs <cluster>
(Press control-C to quit monitoring)
-
Check whether the VMs of the cluster are created on the VMware vCenter.
-
Log into the cluster VMs using SSH to see if they are accessible.
-
For a single-node cluster, log into the -ops
VM. For multinode clusters, log into one of the control plane VMs using SSH with the correct username and the SSH private key file.ssh -i <private-key-file> <cluster-user>@<vm-ip-address>
-
Check the Kubernetes cluster using the kubectl command.
For example, to check the status of all pods, use the following command: kubectl get pod --all-namespaces
When all pods are in the
Running
state, you can log in to the Cisco Smart PHY UI page.