I'm trying to get an EKS cluster to work with the efs csi driver. https://www.pulumi.com/blog/persisting-kubernetes-workloads-with-amazon-efscsi-volumes-using-pulumi-sdks/ this suggests using the helm chart as the other advice there is out of date. But when I tried that I ran into an issue where the driver is already in the cluster and thus pulumi fails to install it, which I wrote up here: https://github.com/kubernetes-sigs/aws-efs-csi-driver/issues/685 but I closed that issue as the helm chart works fine if I use helm to install it and not pulumi, and pulumi gives the hooks warning so I assume the hooks are what is deleting the already existing driver from the cluster so the chart can install a new one (and it does as the timestamp on the driver updated when I installed the chart with helm.
So, how can I fake up what the helm hooks are doing? Is there a way in pulumi for me to delete the csidriver which it seems EKS installs by default? I could obviously do it using kubectl, but it would be nicer if I could capture that in the pulumi code so future me can follow the breadcrumbs
04/28/2022, 6:28 PM
@polite-napkin-90098 based on that issue, looks like you're using
- if you use
04/28/2022, 6:29 PM
oh... thanks I shall take a look at helm.v3.Release, thanks
Yes that seems to have worked without errors, although the StorageClass doesn't seem to have been created. I'll investigate why some more and/or add more code to make it.