OpenShift 4.4 goes all out on mixed workloads, puts observability at devs’ fingertips

OpenShift 4.4 goes all out on mixed workloads, puts observability at devs’ fingertips
Forklift and container, image via Shutterstock

Announced last week, version 4.4 of Red Hat’s container application platform OpenShift is now finally ready to be downloaded, doing its bit to put the dev into DevOps and provide better insight and integration for all sorts of workloads.

At first look, the OpenShift team seems to have been busy making the platform more attractive for a more diverse array of workloads. For example, it comes with a preview for running Windows Server Containers, and KubeVirt-based OpenShift virtualization, which lets users throw VM workloads into the mix as well.

To have more influence on the distribution of all of these loads, version 4.4 includes a descheduler tool for administrators to rebalance workloads across an OpenShift customer. It allows them to evict running pods which can then be rescheduled “onto a more suitable node” to utilise the setup better. Platform admins can also manage the container density on nodes by overriding the ratio between requests and limits set on developer containers via the Cluster Resource Override Operator.

Apart from this, the ingress controller implementation has been reworked to use HAProxy 2.0, which should improve security and performance, and now includes a route admission policy API. OpenShift DNS has learned to “resolve name queries for endpoints using DNS servers outside the cluster” for better integration in hybrid cloud scenarios.

In a bid to not miss the boat on edge computing, the team also made the platform able to work with the stream control transmission protocol (SCTP) which allows for more fault tolerant connections by enabling multiple IP addresses.

In order to use OpenShift in a true DevOps manner, app and service developers need a way of finding out how their workloads are doing without necessarily involving operations. Version 4.4, therefore, finally comes fitted with a metrics and monitoring view as part of the developer console, which provides them with some insight into performance. 

Devs in need of help when building cloud-agnostic serverless applications, can meanwhile draw on OpenShift Serverless once their system has been updated. The Knative-based project can be used to package apps in Linux containers which are meant to run on a variety of cloud platforms, and can be triggered by a number of events. OpenShift Serverless just got promoted into general availability, meaning technical support to get going is available.

Speaking of promotions, the Pipelines feature that was introduced last summer to give teams the tools to create CI/CD workflows and automate application delivery has moved into its Tech Preview phase.

OpenShift 4.4 also contains a preview of Builds which complements Pipelines nicely by providing a way of using Buildah, Source-2-Image and similar tools in the Kubernetes realm to turn application code into container images for easier distribution. Building stateful applications has gotten a bit easier in 4.4 thanks to new resize, snapshot, restore, and clone capabilities for persistent volumes, while Helm releases can now be created from the charts the Developer Catalog provides.

In terms of operators, version 4.4 of the container platform is the first to include a etcd cluster operator, which is useful for disaster recovery since it for example helps to restore previous cluster states or reports events needed for cluster debugging. Another change can be found in the Insights Operator which “periodically collects anonymized certificate signing requests (CSR) to identify CSRs that are not verified in Kubernetes or have not been approved” in v4.4.

Those looking to learn more should take a stroll through the newly updated docs. OKD users meanwhile seem to have to wait just a tad longer, since the distro is still in its fourth beta at the time of writing.