Upgrade from/to RC is not supported.

Longhorn v1.1.0 is a feature release.

Highlights

1. Native RWX support (experimental)
   1. Documentations are coming soon.
2. ARM64 support (experimental)
   1. Longhorn now supports ARM64 architecture in addition to AMD64. You can deploy Longhorn as usual in ARM64 without any modification.
3. CSI Snapshotter support
   1. CSI Snapshotter is a new beta feature since Kubernetes v1.17. It allows users to create snapshots for volumes directly using the kubectl command. In Longhorn's context, those CSI snapshots will be mapped into backups. So after we add the support, Longhorn users will be able to create/restore backups for the volumes without using the Longhorn UI.
4. Prometheus support
   1. We've added Prometheus endpoints and an example Grafana dashboard.
   2. Now the users can easily use the existing monitoring solution to collect stats about the Longhorn, as well as set up alerts for certain situations.

Enhancements

1. Enhanced failure recovery during Kubernetes node failure
   1. Automatic recovery for the StatefulSets.
   2. Improved recovery mechanism for volume failure.
   3. Rebuild based on existing replica after the node temporarily disconnected.
2. Better node maintenance support
   1. Automatically remove node when the node has been removed from Kubernetes.
   2. Automatically recognized the existing disk on another node (e.g. reattached EBS on a different EC2 instance).
   3. Support Kubernetes drain.
   4. Support manually pause replica rebuild for server maintenance.
   5. Support replica eviction for the nodes/disks.
3. Support Data Locality option to always keep a replica local to the engine.
   1. The data locality feature will always try to keep a replica local to the workload, which guarantees even when the network interruption happens, the volume will continue working for the local workload.
4. Allow volume creation with degraded availability
   1. If the value is true, Longhorn will not prevent the volume creation as long as there is at least one replica has been scheduled successfully. It will help users with less than 3 nodes environment. Longhorn will automatically start rebuilding replicas if more available space on other nodes is found later.
   2. The default value is true. Recommend value for the production environment: false.
5. Improvement performance via removing revision counter (experimental).

Release note

1. The minimal required Kubernetes version for Longhorn v1.1.0 is v1.16.
   1. See here for details.
2. The minimal required Kubernetes version for Longhorn with CSI Snapshotter is v1.17.
   1. Please follow the instruction here to install the CSI Snapshot Controller in your Kubernetes cluster if you want to use the CSI snapshot feature.
3. Experimental features:
   1. For the experimental features in this release, we will continue working on improving them. We don't recommend using the experimental features in production.

Upgrade

Live upgrade from v1.0.x is supported. Follow the upgrade instructions here