Install Retries Can Fail Because Cluster Metadata Exists about hive

[dgoodwin]

Admin kubeconfig too.

[joelddiaz]

What do we do in the case where there is an existing configmap/secret? Trash/replace?

[dgoodwin]

I think they fall into the same bucket as the cleanup before install, so yeah I'd trash them if they exist and we're runinng a new install job.

[joelddiaz]

after discussion the plan is to:

1. detect any pre-existing secret/configmap and delete
2. 10 retries to fetch/store the admin kubeconfig (this is considered a critical failure)
3. best effort to save the installer's metadata file
4. update the clusterdeployment status with the secret name holding the admin kubeconfig (non-critical as it will be updated during normal controller reconciliation)

[joelddiaz]

fixed with #92