Skip to main content

Support and troubleshooting

Last updated on

This topic describes the issues you are most likely to hit when you install the Harness SMP Operator, and how to resolve them.

Common installation issues​

Each entry names the usual cause, gives a command that confirms it, and states the fix. Run the Check command first to confirm that you are looking at the right problem, then apply the Fix.

Operator pods stuck in ImagePullBackOff​

  • Cause: Images are not accessible from the cluster (wrong registry, missing credentials, or images not mirrored).
  • Check: kubectl describe pod -n harness platform-installer-0 | grep -A5 Events
  • Fix: Verify registry credentials and ensure all required images are pushed. Go to Registry setup to review the registry requirements.

DNS not resolving​

  • Cause: DNS record not pointing to the load balancer or ingress external IP.
  • Check: nslookup <YOUR_DNS>
  • Fix: Create an A or CNAME record pointing to the external IP of your ingress service (kubectl get svc -n harness).

Platform PVCs stuck in Pending​

  • Cause: Storage class does not exist or no dynamic provisioner is available.
  • Check: kubectl get pvc -n harness
  • Fix: Verify the storage class exists (kubectl get storageclass) and matches the value passed through --pi-set cluster.tfi.storageClass=<name>. Go to Storage classes to review the supported storage classes.

Installer UI not accessible​

  • Cause: Ingress or VirtualService not configured correctly, or TLS certificate issue.
  • Check: kubectl get ingress -n harness (nginx) or kubectl get vs -n harness (Istio)
  • Fix: Verify the ingress resource exists and the external IP is reachable. Check the TLS secret with kubectl get secret -n harness | grep tls.

Get support​

Contact Harness Support if the entries above do not resolve your issue.

  • Prerequisites: Cluster, storage, registry, and DNS requirements to confirm before you install.
  • Post-installation checks: Verify that each part of the deployment came up correctly.
  • CLI reference: Installation parameters for the helm and clustermgr flows.