Skip to main content

Release Announcement v1.1.0

We are pleased to announce that the Apache YuniKorn community has voted to release 1.1.0. Apache YuniKorn is a standalone resource scheduler, designed for managing, and scheduling Big Data workloads on container orchestration frameworks like Kubernetes for on-prem and on-cloud use cases.

Overview

The Apache YuniKorn community has fixed 89 JIRAs in this release.

Release manager: Peter Bacsko

Release date: 2022-09-08

Highlights

REST API documentation and enhancements

The REST API now can return the details of a specific application and list the pending allocations of an application. Documentation of the REST API have also been enhanced.

Multi-architecture build

With the ARM architecture becoming more popular, we now build binaries and Docker images for both amd64 and arm64 targets.

Recovery stabilization

Several issues have been identified during the recovery phase of Yunikorn which mostly affected gang scheduling (eg. running placeholders getting replaced immediately) but also Spark workloads.

DaemonSet scheduling

Scheduling of DaemonSet pods were problematic before this release. If the node was full, then those pods might not have been scheduled. However, it is usually important to start DeamonSet pods as they often perform tasks that are necessary on all nodes like log collection and aggregation, resource monitoring, storage management, etc. meaning they have priority over regular application pods.

The implementation of YUNIKORN-1085 ensures that we have a predictable preemption mechanism which terminates running pods if necessary to make room for DaemonSet pods.

e2e testing improvements

Additional end-to-end tests have been written to increase the coverage of Yunikorn as we support more K8s versions.

Community

The Apache YuniKorn community is pleased to welcome new PMC members Peter Bacsko, Manikandan Ramaraj and committer Ted Lin.