This usually happens when the state of the resources in the Kubernetes cluster does not match what Helm knows about. Here the namespace had been deleted without first deleting the helm releases. So helm attempted to perform an upgrade on a non-existent resource.
To fix this, simply delete the errant helm release. In this case, we would run the following command to delete the pr-1627-app helm release.
$ helm delete --purge pr-1627-app
I hope this will help.