> For the complete documentation index, see [llms.txt](https://developer.harness.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://developer.harness.io/continuous-delivery/troubleshooting-and-resources/armory/general/deployment-strategy-red-black-error-load-balancer-service-service-name-does-not-exist.md).

# Deployment Strategy Red/Black error- Load balancer service does not exist

### Issue <a href="#issue" id="issue"></a>

With the Deploy Strategy set to ***Red/Black***, and when deploying to Kubernetes using ***Kustomize***, the bake stage is successful and the service section is visible in the baked manifest. However the pipeline fails with the following error: `Load balancer service does not exist` The UI shows the following, for example:

### Cause <a href="#cause" id="cause"></a>

This issue is related to the following Github issues:<https://github.com/spinnaker/spinnaker/issues/5040><https://github.com/spinnaker/clouddriver/pull/5248> As reported, Kubernetes has a bug where it will not see the manifest properly and will therefore error out. When utilizing the `Deploy (Manifest)` stage with ***Red/Black rollout strategy***, a requirement of using this stage was for a `Service` to exist before running the stage in order to read its `selector` and apply it to the newly deployed replicaset. This requirement has been updated, so that now the `Service` will first be read from the stage manifest list, and if not found it will proceed with the same behaviour as before, reading it live from the cluster.