Severity
Sev-3, blocking installation on some Helm versions. Workarounds available.
Version
Reported on 2026.2.8245
Latest Version
Not applicable
What happened?
A customer encountered a number of issues with the Helm command that gets generated by the Kubernetes Installation Wizard.
--atomic is not a valid option for Helm versions [4.0.0 - 4.1.3).
- NFS is disabled by default from Helm 3.0.1
Reproduction
Navigate to the Kubernetes Agent installation wizard in Octopus Deploy and follow the generated Helm command on a Helm 4 cluster, or on a cluster without a pre-configured storage class.
More Information
As of 3.0.1, NFS support is off by default. The wizard should guide users through storage class configuration and include prompts for enabling NFS where needed. With the planned deprecation of the bundled NFS option, the wizard should also help users prepare their cluster with a supported storage class.
Workaround
Modify the wizard-generated command before running on your cluster.
If your helm instance does not support --atomic
- Replace
--atomic with --rollback-on-failure
If your cluster does not have a pre-configured ReadWriteMany-capable storage class:
- Add
--set persistence.nfs.enabled="true" and --set 'persistence.accessModes={ReadWriteMany}'
Severity
Sev-3, blocking installation on some Helm versions. Workarounds available.
Version
Reported on 2026.2.8245
Latest Version
Not applicable
What happened?
A customer encountered a number of issues with the Helm command that gets generated by the Kubernetes Installation Wizard.
--atomicis not a valid option for Helm versions[4.0.0 - 4.1.3).Reproduction
Navigate to the Kubernetes Agent installation wizard in Octopus Deploy and follow the generated Helm command on a Helm 4 cluster, or on a cluster without a pre-configured storage class.
More Information
As of 3.0.1, NFS support is off by default. The wizard should guide users through storage class configuration and include prompts for enabling NFS where needed. With the planned deprecation of the bundled NFS option, the wizard should also help users prepare their cluster with a supported storage class.
Workaround
Modify the wizard-generated command before running on your cluster.
If your helm instance does not support
--atomic--atomicwith--rollback-on-failureIf your cluster does not have a pre-configured
ReadWriteMany-capable storage class:--set persistence.nfs.enabled="true"and--set 'persistence.accessModes={ReadWriteMany}'