Releases: NVIDIA/KAI-Scheduler
Releases · NVIDIA/KAI-Scheduler
v0.9.3
What's Changed
- Update the ray grouper plugin by @davidLif in #507
- feat: add prometheus operand by @SiorMeir in #477
- Time fairshare - support tumbling window config by @davidLif in #490
- Fix podgroup condition update by @itsomri in #485
- NodeSet plugin infra by @omer-dayan in #496
- Add Scheduler Concepts to developer documentation by @gshaibi in #509
- adding scheduler operand by @enoodle in #493
- adding operator integration tests by @enoodle in #510
- Added more docs about workload dedicated namespaces by @romanbaron in #516
New Contributors
Full Changelog: v0.9.2...v0.9.3
v0.9.2
What's Changed
- kai operator docs by @enoodle in #484
- allow providing scheme to podgrouper app by @natasharomm in #501
- Design doc - NodeSet feature by @omer-dayan in #491
- operator to deploy binder by @enoodle in #495
- Grove: Add support for PodGangSet to PodCliqueSet API name change by @sanjaychatterjee in #508
Full Changelog: v0.9.1...v0.9.2
v0.9.1
What's Changed
- allow providing scheme to podgrouper app by @natasharomm in #502
Full Changelog: v0.9.0...v0.9.1
v0.9.0
What's Changed
- Added details to waitForPGConditionReason in e2e by @itsomri in #470
- add queue controller operand by @enoodle in #469
- deploy operator in chart with pod group controller by @enoodle in #467
- Topology scheduling scheduler tests by @omer-dayan in #476
- Time aware fairness: prometheus time decay by @itsomri in #462
- [WIP]: Added time aware fairness proposal by @itsomri in #311
- add scale adjuster operand by @enoodle in #473
- Align ginkgo versions by @itsomri in #478
- operator deploy admission queue controller by @enoodle in #474
- Topology scheduling tests by @omer-dayan in #479
- add pod grouper operand by @enoodle in #475
- operator deploy pod grouper by @enoodle in #483
- cleanup scheduler args by @enoodle in #487
- using group+kind to get default priority and category from configmap by @natasharomm in #488
- adding binder operand by @enoodle in #486
- introduce scheduling shards by @enoodle in #482
- deploy node scale adjuster from kai operator by @enoodle in #492
- Bugfix - Prefer topology level under root by @omer-dayan in #480
- Consider historical usage data in fair share calculations by @davidLif in #494
- Fix changelog for v0.9.0 by @itsomri in #498
Full Changelog: v0.8.5...v0.9.0
v0.8.5
What's Changed
- PGInfo Clean SubGroup messages + Remove GetDefaultMinAvailable by @omer-dayan in #454
- Add scheduler topology envtest by @davidLif in #449
- adding initial KAI-operator by @enoodle in #459
- adding admission operand by @enoodle in #465
- If a podgroup has scheduled pod and a required topology constraint, t… by @davidLif in #460
- added configurable plugins hub for podgrouper using interface and RegisterPlugins by @natasharomm in #468
Full Changelog: v0.8.4...v0.8.5
v0.8.4
What's Changed
- Bump k8s.io/kubernetes from 1.32.6 to 1.32.7 by @dependabot[bot] in #448
- Add job order reflection plugin with HTTP endpoint by @singh1203 in #445
- using values from constants.go in options by @enoodle in #455
- Fixed a bug where workload with subgroups would not consider addition… by @romanbaron in #457
- move crds into helm chart by @enoodle in #451
Full Changelog: v0.8.3...v0.8.4
v0.8.3
What's Changed
- Removed gpu operator as dependency by @itsomri in #438
- Roman/subgroups e2e by @romanbaron in #434
- Basic topology e2e test by @davidLif in #412
- Roman/subgroup pod status cache by @romanbaron in #429
- Few renames in topology_plugin_job_filtering.go by @romanbaron in #366
- Add pods to Default subGroup only if not part of subGroups by @omer-dayan in #408
- Time aware fairness usage: prometheus client by @itsomri in #435
- Not using summerized MinAvailable for PodGroup, but look per SubGroup by @omer-dayan in #443
- removing unused arguments from admission by @enoodle in #450
- remove unused variable and volumes from binder by @enoodle in #452
- Fix changelog v0.8.3 by @itsomri in #453
Full Changelog: v0.8.2...v0.8.3
v0.6.14
v0.4.15
v0.8.2
What's Changed
- IsInferencePreemptible cleanup by @romanbaron in #420
- fix: status updater deleted pod group race by @enoodle in #421
- Inject build info from makefile by @itsomri in #428
- fix: GPU memory calculation using incorrect unit conversion by @eveningcafe in #417
- Added usage parameters to proportion plugin by @itsomri in #418
- Fix: incorrect scheduling decision and calculation when using MIG by @hello2mao in #422
- update changelog by @enoodle in #431
- Time aware fairness client configs by @itsomri in #419
- Update gpu operator import by @itsomri in #433
- Fixed changelog v0.8.2 by @itsomri in #436
New Contributors
- @eveningcafe made their first contribution in #417
Full Changelog: v0.8.1...v0.8.2