Partially-installed Releases are never retried

[nojnhuh]

What steps did you take and what happened:

I haven't found a reliable way to reproduce this yet, but I observed CAAPH hit a transient failure when installing a chart:

E0729 01:35:05.341942       1 controller.go:324] "Reconciler error" err="Kubernetes cluster unreachable: Get \"https://capz-0bei2n-6bb3419a.canadacentral.cloudapp.azure.com:6443/version?timeout=10s\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)" controller="helmreleaseproxy" controllerGroup="addons.cluster.x-k8s.io" controllerKind="HelmReleaseProxy" HelmReleaseProxy="default/azuredisk-csi-driver-capz-0bei2n-c4phf" namespace="default" name="azuredisk-csi-driver-capz-0bei2n-c4phf" reconcileID="f99b16ad-79e0-41d4-a02a-d580336d4fe8"

This left the Helm Release in a pending-install state. Then in the next and all future reconciliations of the HelmReleaseProxy, CAAPH identified the release as "up to date" and did not reconcile it further, so the resources associated with the Release never finished installing:

I0729 01:35:15.083876       1 helm_client.go:370] "Release `azuredisk-csi-driver-oot` is up to date, no upgrade required, revision = 1" controller="helmreleaseproxy" controllerGroup="addons.cluster.x-k8s.io" controllerKind="HelmReleaseProxy" HelmReleaseProxy="default/azuredisk-csi-driver-capz-0bei2n-c4phf" namespace="default" name="azuredisk-csi-driver-capz-0bei2n-c4phf" reconcileID="ee9e4ebb-243f-421c-b0d7-08a77c30f577"

What did you expect to happen:

CAAPH tries to continue the install.

Anything else you would like to add:

The full CAAPH logs come from this CAPZ load test run.

Environment:

• Cluster API version: v1.10.4
• Cluster API Add-on Provider for Helm version: v0.2.5
• minikube/kind version:
• Kubernetes version: (use kubectl version): v1.34.0-beta.0.567+dd4e4f1dd13f68
• OS (e.g. from /etc/os-release):

/kind bug