v0.4.8

Fixed

• Queue order function now takes into account potential victims, resulting in better reclaim scenarios.

CHANGED

• Cached GetDeservedShare and GetFairShare function in the scheduler PodGroupInfo to improve performance.
• Added cache to the binder resource reservation client.
• More Caching and improvements to PodGroupInfo class.
• Update pod labels after scheduling decision concurrently in the background.

v0.5.0

What's Changed

• Fix typo in priority docs by @Sovietaced in #108
• Remove repeated active departments queue creation by @bgedik in #113
• Fix filename for stalegangeviction by @ArmedGuy in #115
• [Bugfix] Prep for queue comparison changes by @itsomri in #114
• Add/amend test logging to use .TestTopologyBasic.Name by @ArmedGuy in #116
• Made resource reservation parameters configurable by @romanbaron in #106
• Add Changelog to track changes by @itsomri in #117
• Changed to regular ubuntu docker image in tests by @romanbaron in #123
• Cluster autoscaler adjustment for GPU sharing pods by @romanbaron in #119
• added contributing, maintainer and owners files by @romanbaron in #74

New Contributors

• @Sovietaced made their first contribution in #108
• @bgedik made their first contribution in #113
• @ArmedGuy made their first contribution in #115

Full Changelog: v0.4.7...v0.5.0

v0.4.7

What's Changed

• fix: snapshot tool cache.Run call by @enoodle in #102
• Docs hotfix: Update and rename pytorch-elasitc.yaml to pytorch-elastic.yaml by @EkinKarabulut in #105
• Adding Issue Templates for bug & feature/enhancement requests by @EkinKarabulut in #103
• fix: gpu resource device count calculation by @enoodle in #107

Full Changelog: v0.4.6...v0.4.7

v0.4.6

What's Changed

• Initialize metrics namespace on scheduler run by @romanbaron in #100

Full Changelog: v0.4.5...v0.4.6

v0.4.5

What's Changed

• fix transition id comparison by @enoodle in #87

Full Changelog: v0.4.4...v0.4.5

v0.4.4

What's Changed

• Removed unneeded runai references in the code by @romanbaron in #83
• Bump k8s.io/kubernetes from 1.32.1 to 1.32.4 by @dependabot in #85
• Making scheduler metrics subsystem configurable by @romanbaron in #86

Full Changelog: v0.4.3...v0.4.4

v0.4.3

What's Changed

• Fixed package version expression resolution by @romanbaron in #80

Full Changelog: v0.4.2...v0.4.3

v0.4.2

What's Changed

• Add delayedLauncherCreationPolicy handeling to the pod-grouper by @davidLif in #75
• Bump golang.org/x/net from 0.37.0 to 0.38.0 by @dependabot in #76
• docs: specify legacy v1 version of kubeflow-training-operator by @tgasla in #71
• Add a simple DRA example by @guptaNswati in #59
• BUG fix - handle pod grouper top owner climbing RBAC error by @davidLif in #77
• Changed nvcr to ghcr in all workflows, scripts and docs by @romanbaron in #78

New Contributors

• @dependabot made their first contribution in #76

Full Changelog: v0.4.1...v0.4.2

v0.4.1

What's Changed

• Adding support channels to README.md by @EkinKarabulut in #64
• storing CI artifacts on ghcr.io by @romanbaron in #58
• Main- Bug fix - when syncing the inFlight PG status, override only if the t… by @davidLif in #70
• fix(status-updater): memory leak fix - remove unused inFlight PGs by @enoodle in #73

New Contributors

• @EkinKarabulut made their first contribution in #64

Full Changelog: v0.4.0...v0.4.1

v0.4.0

What's Changed

• chore: improve pr ci time by @enoodle in #55
• e2e- knative integration by @davidLif in #56
• E2E - restrict node scheduling by @davidLif in #57
• Bug fix - handle fractions in status.queueDetails correctly by @davidLif in #61

Full Changelog: v0.3.0...v0.4.0