During an upgrade, Ingress errors are seen and the installation stops

Problem

The installation logs contain a number of references to Ingress and appear connected to certain helm charts

Resolution

Identify the helm charts by running a describe on the ingress which fail, i.e.

$ kubectl -n domino-platform describe ingress | grep helm
...
meta.helm.sh/release-namespace: domino-platform
meta.helm.sh/release-name: pusher-service
meta.helm.sh/release-namespace: domino-platform
meta.helm.sh/release-name: web-ui
meta.helm.sh/release-namespace: domino-platform

From your bastion or installation node, run helm list, you should be able to associate the results from the installer, kubectl output to the helm charts in your list.

Use the helm uninstall command to remove those charts and then retry your installation.

From your installation log you can also see the old vs new and confirm you are removing the 'old' version.

Alternatively install this helm plugin: https://github.com/helm/helm-mapkubeapis

Root cause

The method of tagging helm can cause issues on self-installed and sandboxed Domino Deployments and failed attempts to upgrade. The prior API release may be deleted on the first upgrade attempt, with the recourse listed above, to delete those items manually on the retry.