The British company Canonical, which is best known for the Linux distribution Ubuntu, has introduced a high availability function for its slim Kubernetes variant MikroK8s. It needs at least three nodes and is supposed to provide fail-safety. If a node malfunctions, the cluster heals itself.

MicroK8s is primarily aimed at development workstations and use at the edge or on the Internet Things. The short form K8s, in which the number replaces the eight letters “ubernete”, is also used by other K8s providers. The lean distribution is available as an open source project on GitHub. It comes as a single package and aims at low resource requirements and simple administration.

You should be three nodes The newly introduced high availability should also allow the lean clusters to be operated in a fail-safe manner. Specifically, the cluster should continue to run smoothly if a component fails. To do this, the MicroK8s cluster must have at least three nodes. In this case the Dqlite datastore, which manages the cluster status, is automatically highly available.

In-depth lectures on Kubernetes and the entire ecosystem are available on the two Kubernetes theme days as part of the Continuous Lifecycle and ContainerConf 2020 :

4. November 2020: Kubernetes Professionals Day 20. January 2021: Kubernetes Experts Day Dqlite is a special variant of SQLite, and the name stands for Distributed SQLite. The datastore is designed for distributed applications and offers automatic failover by default. It relies on the Raft Consensus Algorithm, which was developed to manage fault tolerance as an alternative to the Paxos protocols. In the case of clusters with more than three nodes, the system uses the additional nodes as a reserve and activates them in the event of a failure