Sylva-units periodic reconciliation in (capm3?) bootstrap cluster

Summary

Sylva-units periodic reconciliation in capm3 bootstrap cluster is detecting drift after 30 minutes and generating a new release revision with a new root dependency.

This has been observed while working on #3907 resolution, and has been observed in almost all referred pipelines:

https://gitlab.com/sylva-projects/sylva-core/-/jobs/13843937302 https://gitlab.com/sylva-projects/sylva-core/-/jobs/13885916639 https://gitlab.com/sylva-projects/sylva-core/-/jobs/13886264938

We can see that sylva-units HelmRelease in bootstrap cluster is reconciling 2 times:

    history:
    - appVersion: 0.0.0
      chartName: sylva-units
      chartVersion: 0.0.0-git-c2cf24ef+9675be9d033a
      configDigest: sha256:43cf5e9b556fd11c8bb7342f598dd61ba98ffd18cbf50df883694adc499725d0
      digest: sha256:59dda1a723d4cf7fe26f91df990ad9ca4da55f2512aa1e1a93b17e51fc8153fb
      firstDeployed: "2026-04-09T09:05:53Z"
      lastDeployed: "2026-04-09T09:36:40Z"
      name: sylva-units
      namespace: sylva-system
      ociDigest: sha256:9675be9d033a8f846eb876dbcba6b3f025eb18c2b904ce7e59941a1e26ae3e93
      status: deployed
      version: 2
    - appVersion: 0.0.0
      chartName: sylva-units
      chartVersion: 0.0.0-git-c2cf24ef+9675be9d033a
      configDigest: sha256:28d231e0522bee9184f69cc498dcbf969e04dd4e938cc5d49d92d744c9d432d2
      digest: sha256:3391ab5616b23ae001784a5623b72b57ebcbd72692f40d8907b39272c9ac5f7e
      firstDeployed: "2026-04-09T09:05:53Z"
      lastDeployed: "2026-04-09T09:05:53Z"
      name: sylva-units
      namespace: sylva-system
      ociDigest: sha256:9675be9d033a8f846eb876dbcba6b3f025eb18c2b904ce7e59941a1e26ae3e93
      status: superseded
      version: 1

In helm controller logs, we can see that the controller has seen some config values changes:

{"level":"info","ts":"2026-04-09T09:36:26.007Z","msg":"release out-of-sync with desired state: release config values changed","controller":"helmrelease","controllerGroup":"helm.toolkit.fluxcd.io","controllerKind":"HelmRelease","HelmRelease":{"name":"sylva-units","namespace":"sylva-system"},"namespace":"sylva-system","name":"sylva-units","reconcileID":"05169c2c-13d1-4ffd-800b-4df0934ca3e5"}
{"level":"info","ts":"2026-04-09T09:36:26.026Z","msg":"running 'upgrade' action with timeout of 5m0s","controller":"helmrelease","controllerGroup":"helm.toolkit.fluxcd.io","controllerKind":"HelmRelease","HelmRelease":{"name":"sylva-units","namespace":"sylva-system"},"namespace":"sylva-system","name":"sylva-units","reconcileID":"05169c2c-13d1-4ffd-800b-4df0934ca3e5"}
{"level":"info","ts":"2026-04-09T09:36:48.550Z","msg":"release in-sync with desired state","controller":"helmrelease","controllerGroup":"helm.toolkit.fluxcd.io","controllerKind":"HelmRelease","HelmRelease":{"name":"sylva-units","namespace":"sylva-system"},"namespace":"sylva-system","name":"sylva-units","reconcileID":"05169c2c-13d1-4ffd-800b-4df0934ca3e5"}

Consequently all the generated units are reconciled a second time with new root dependency, which can possibly cause extra delays before the pivot.

First option would obviously be to increase the reconciliation interval (it was set to 24h before the introduction of sylva-units-operator in sylva-1.6).

But we should ideally understand which changes are seen by the controller, and prevent this drift correction.