Here we cover how to upgrade to the latest Longhorn from all previous releases.

Deprecation & Incompatibility

There are no deprecated or incompatible changes introduced in v1.5.3.

Upgrade Path Enforcement

Since Longhorn v1.5.0, Longhorn only allows upgrades from supported versions. If upgrading from any unsupported version, the upgrade will fail. However, users can revert to the previous state without any service interruption or downtime.

In addition, Longhorn disallows downgrading to any previous version to prevent unexpected system statuses caused by potential function incompatibility, deprecation, or removal. Please refer to the following matrix to understand the supported upgrade versions and there is an example for each upgrade path:

Warning:

• Upgrade path enforcement is introduced in Longhorn v1.5.0, which means that downgrading from v1.5.0 to any previous version is possible. Please note that downgrading is not supported.

Upgrading Longhorn

There are normally two steps in the upgrade process: first upgrade Longhorn manager to the latest version, then manually upgrade the Longhorn engine to the latest version using the latest Longhorn manager.

1. Upgrade Longhorn manager

• To upgrade from v1.5.x (< v1.5.3) or v1.4.x, see this section.

2. Manually Upgrade Longhorn Engine

After Longhorn Manager is upgraded, Longhorn Engine also needs to be upgraded using the Longhorn UI.

3. Automatically Upgrade Longhorn Engine

Since Longhorn v1.1.1, we provide an option to help you automatically upgrade engines

4. Automatically Migrate Recurring Jobs

With the introduction of the new label-driven Recurring Job feature, Longhorn has removed the RecurringJobs field in the Volume Spec and planned to deprecate RecurringJobs in the StorageClass.

During the upgrade, Longhorn will automatically:

• Create new recurring job CRs from the recurringJobs field in Volume Spec and convert them to the volume labels.
• Create new recurring job CRs from the recurringJobs in the StorageClass and convert them to the new recurringJobSelector parameter.

Visit Recurring Snapshots and Backups for more information about the new Recurring Job feature.

5. Detach All V1 and V2 Volumes When v2-data-engine is Enabled Before Upgrade

Longhorn does not support upgrading with v1 or v2 volumes attached when the V2 Data Engine is enabled. Detach all volumes before upgrading.

Extended Reading

Visit Some old instance manager pods are still running after upgrade for more information about the cleanup strategy of instance manager pods during upgrade.

Need Help?

If you have any issues, please report it at https://github.com/longhorn/longhorn/issues and include your backup yaml files as well as manager logs.