English | 简体中文

What is NEW!
May 19th, 2020. Kruise v0.5.0 is RELEASED! It supports maxSurge for CloneSet and fixes bugs for StatefulSet/SidecarSet, please check the CHANGELOG for details.
Mar 20th, 2020. Kruise v0.4.1 is RELEASED! It provides graceful in-place update for Advanced StatefulSet and CloneSet, please check the CHANGELOG for details.
Nov 24th, 2019. A blog about new UnitedDeployment controller is posted in Kruise Blog (link).

Kruise is the core of the OpenKruise (official site: https://openkruise.io) project. It consists of several controllers which extend and complement the Kubernetes core controllers for workload management.

As of now, Kruise offers five workload controllers:

• CloneSet: CloneSet is a workload that mainly focuses on managing stateless applications. It provides a rich set of features for more efficient, deterministic and controlled management, such as in-place update, specified Pod deletion, configurable priority/scatter based update, preUpdate/postUpdate hooks, etc. This post provides more details about why CloneSet is useful.

• Advanced StatefulSet: An enhanced version of default StatefulSet with extra functionalities such as in-place update, pause and maxUnavailable.

• SidecarSet: A controller that injects sidecar containers into the Pod spec based on the Pod selectors. The controller is also responsible for upgrading the sidecar containers.

• UnitedDeployment: This controller manages application Pods spread in multiple fault domains by using multiple workloads.

• BroadcastJob: A job that runs Pods to completion across all the nodes in the cluster.

The project roadmap is actively updated in here. This video demo by Lachlan Evenson is a good introduction for new users.

• In-place update

  In-place update provides an alternative to update container images without deleting and recreating the Pod. It is much faster compared to the recreate update used by the native Deployment/StatefulSet and has almost no side effects on other running containers.

• Sidecar containers management

  The Sidecar containers can be simply defined in the SidecarSet custom resource and the controller will inject them into all Pods matched. The implementation is done by using Kubernetes mutating webhooks, similar to what istio does. However, SidecarSet allows you to explicitly manage your own sidecars.

• Multiple fault domains deployment

  A global workload can be defined over multiple fault domains, and the Kruise controller will spread a sub workload in each domain. You can manage the domain replicas, sub workload template and update strategies uniformly using the global workload.

For a Kubernetes cluster with its version higher than v1.12, you can simply install Kruise with helm v3:

helm install kruise https://github.com/openkruise/kruise/releases/download/v0.5.0/kruise-chart.tgz

Note that installing this chart directly means it will use the default template values for the kruise-manager. You may have to set your specific configurations when it is deployed into a production cluster or you want to enable/disable specific controllers.

For more details, see quick-start.

You can view the full documentation from the OpenKruise website.

We also provider tutorials to demonstrate how to use Kruise controllers.

You are warmly welcomed to hack on Kruise. We have prepared a detailed guide CONTRIBUTING.md.

Active communication channels:

• Slack: channel address
• Mailing List: todo
• Dingtalk Group(钉钉讨论群)

Kruise is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.