OpenShift boards the GitOps train

OpenShift boards the GitOps train

OpenShift 4.7 has landed, aligning the platform with various other Kubernetes distributions that are slowly catching up with the K8s 1.20 release from last December and throwing in a couple of technical previews for extra excitement.

OpenShift’s web console is now available with localisation for Chinese, Japanese, and Korean speaking areas. It also features an API Performance dashboard which can be used to inspect API server metrics such as request rate and duration, response status codes, and the number of requests aborted. Speaking of monitoring, v.4.7 comes with namespace and pod Kubernetes networking dashboards in Grafana enabled, and a whole array of changed alerting rules, so a brief look into the documentation is advised. 

With container experts continuing to show interest in the GitOps approach, Red Hat also decided to demonstrate appreciation by adding a tech preview for setting up continuous deployment in a declarative manner. GitOps usually describes a concept for cluster management and application delivery centring around version control tool git as a single source of truth. 

OpenShift’s implementation gets the necessary functionality through open source projects Argo CD and Tekton Pipelines, though it is worth noting that Tekton hasn’t crossed the 1.0 line yet. While not bad per se, this means that breaking changes could be introduced any time, so that it might still take a while for the GitOps feature to stabilise.

Besides stability, security is a main concern of the standard OpenShift clientele, so the OpenShift Compliance Operator has seen a couple additional checks inspired by the CIS Kubernetes benchmark, while OVN-IPsec now lets users encrypl OpenShift traffic with IPsec when using Open Virtual Network.

To turn OpenShift into more of an option for bare metal deployments, the team also released previews for an assisted installer, meant to simplify automation in such environments, and scheduling profiles for customising scheduler behaviour to better fit the needs of special use cases. Features that have matured enough to be used in production since the last release include the horizontal pod autoscaler for scaling replication controllers and app deployments based on memory metrics, and the descheduler for evicting pods.
Under the hood OpenShift 4.7 is using Kubernetes 1.20 with cri-o 1.20. Details on removed and deprecated features can be found on the project’s website.