OpenShift 4 Developer Preview

1–30 of 170
OpenShift 4 represents a significant rethink of how a Kubernetes distribution is built, run, and managed. 

In our developer preview we're still trying to nail down all of the rough edges, but the core concepts are:

  • Day 1: A simple, easy to use install on all clouds and on bare metal
    • The installer provisions infrastructure and then brings up the cluster, which takes over and creates and manages nodes
    • A limited set of up front options, because all configuration is managed on cluster post-install
    • Boot Red Hat CoreOS instances, which are fully immutable instances based on RHEL 7 (and the upcoming RHEL 8) kernel with built in Kubernetes node software
      • CRI-O is the default runtime, offering high performance and high reliability with the deeper cluster security integration possible through that project
      • Every host is treated as a replaceable unit, where direct SSH access can be considered a last resort option when something misbehaves
  • Day 2: Manage the cluster as just another API
    • Every important configuration choice is presented as an API, including security, tuning, and extension management, allowing config management the Kubernetes Way
    • All core components are managed by operators, which manage rollout and updates, handle recovery, and reconcile changes in your config as necessary
    • All host management is done through the machine API, which can manage the cloud infrastructure to spin up or replace faulty machines, autoscale, and self-monitor
    • Deliver configuration and OS updates to the host together via the machine config operator, Ignition, and Red Hat CoreOS to transparently manage host software
    • Updating a cluster is as easy as point and click - new updates are offered over the air and a single administrative action can roll out or cancel an upgrade.
      • Future previews will add automated updates and enhanced status reporting on cluster updates
  • Observability of the cluster
    • Continuing improvements from OpenShift 3, the Prometheus-based monitoring stack now integrates into Kubernetes as the default metrics API and offers immediate feedback about the resource consumption of the cluster to end users and administrators
    • Future previews will add the ability to expose deeper Prometheus cluster metrics to tenants on a per namespace, giving you the right mix of control and depth to empower your developers
    • Improvements to alerting and reporting for continuing to improve our top notch experience around insight into your cluster
    • Optional integrated telemetry provides Red Hat with feedback on rollouts of new software, including new versions of the OS, Kubernetes, and the components on top that allow us to ensure that updates hit test and QA systems first for soak in each environment, then when all signals are green allow those updates to become available to production systems.
  • Developers
    • All of the easy to use tools from OpenShift 3 continue forward, with even simpler management experiences for the integrated registry and cluster ingress
    • Builds on the cluster are now run in a limited-privilege container, needing no external access to a Docker daemon thanks to integration with the buildah project
This is only a tiny sample of the changes included in OpenShift 4 - you'll see more details over the next few months as we add documentation and user experiences around those features.  

Please ask questions, give feedback and honest opinions, and help us identify and squash bugs.  If you have questions please check the pinned Frequently Asked Questions post and search for other reports from similar symptoms.