This documentation is for Flux (v1) and Helm Operator (v1). Both projects are in maintenance mode and will soon reach end-of-life. We strongly recommend you familiarise yourself with the newest Flux and start looking at your migration path.
For documentation regarding the latest Flux, please refer to this section.
Reconciliation and upgrades
Now that you know the ins and outs of configuring a release, we are going to have a better look at how the Helm Operator performs the actual Helm release.
On the creation and update of a
HelmRelease the resource is queued for
reconciliation. Besides this all
HelmRelease resources handled by the Helm
operator instance are also queued for reconciliation every
--charts-sync-interval (defaults to 3
Once the queued resource has been picked up by a worker, the Helm Operator attempts to receive the chart for the resource and performs several safe guard checks; if those do not result in an error or instruct to return early, the Helm installation or upgrade is performed.
What triggers an upgrade
To prevent spurious upgrades from happening the Helm Operator performs several safe guard checks before performing the actual upgrade. Bellow you will find an overview of the checks it performs, and what effect they have.
If any of the following equals to
false, the sync process will exit with an
error and no upgrade will be performed:
- The values composed of the merged
.valuesare valid YAML.
- The resources of the Helm release are
marked as being managed by the
- The current state of the Helm release is
The first of the following that equals to
true will result in an upgrade
being performed, otherwise no action is taken:
- No Helm release exists in the Helm storage for the
- This generation of the
HelmReleasehas not been processed before — the generation changes for example when the
- The result of a dry-run upgrade for the
HelmReleasediffers from the latest release in the Helm storage.
NoteMutations to live cluster-state are not detected and thus not reverted. This will however be added in the foreseeable future.
When an upgrade fails, the Helm Operator will stop performing upgrades for the
release as it cannot assume this is a safe procedure, nor does it
a rollback. Instead it will start logging
warnings about the
failed status of the release.
Recovering from this is possible, after having inspected the state of the
release, by getting the Helm release manually in a
deployed state, for
example by performing a rollback or upgrade for the release using
helm rollback <release name>
The antecedent annotation
Right after the Helm Operator performs a Helm release for the
HelmRelease resource, all resources that were the result of this release
are annotated with the antecedent annotation
the value of the annotation equals to
The purpose of this annotation is to indicate that the cause of that resource
HelmRelease. It also functions as a safe guard during reconciliation
to ensure the release is only managed by a single
HelmRelease, as it is
possible that due to a misconfiguration multiple
HelmRelease resources exist
with the same