# kubernetes-sigs/kueue

Kubernetes-native Job Queueing

Repository: https://github.com/kubernetes-sigs/kueue
Canonical: https://ross.abutalabs.com/products/kueue
Homepage: https://kueue.sigs.k8s.io
Language: Go
License: Apache-2.0
License Family: permissive
Topics: k8s-sig-scheduling, k8s, kubernetes
Last push: 2026-08-26T21:07:48+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 86, longevity 100
- inputs: {"age_days": 1659, "days_push": 7, "days_rel": 13, "gap_med": 0, "n_releases_24m": 96}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2906, forks 766 (observed 2026-08-28T04:07:29.082461+00:00)

## What it is
Kueue is a Kubernetes-native job queueing system providing APIs and controllers that decide when batch, HPC, and AI/ML workloads should be admitted to start and when they should be stopped. It manages quotas, priorities, preemption, and dynamic capacity sharing across teams and clusters without replacing the core Kubernetes scheduler.

## Use cases
- queue and prioritize batch jobs on kubernetes
- manage GPU quotas across teams sharing a cluster
- run ML training jobs with fair resource sharing
- preempt low-priority jobs when capacity runs out
- schedule LLM training workloads with topology-aware placement
- dispatch jobs across multiple kubernetes clusters
- borrow idle capacity between team resource pools
- wait for cluster autoscaler to provision nodes before admitting jobs

## When to choose
- you run batch, HPC, or AI/ML workloads on Kubernetes and need quota-based admission control
- multiple teams share a cluster and need fair, preemption-aware resource sharing
- you use Kubeflow, Ray, JobSet, or LeaderWorkerSet and need queueing for those job types
- you want to maximize GPU/CPU utilization by borrowing idle capacity across cohorts
- you need multi-cluster job dispatching from a central control plane

## When to avoid
- you only run a handful of long-lived services rather than batch jobs
- you need full batch-scheduler features like gang scheduling with complex dependencies that a dedicated HPC scheduler provides
- your cluster is older than Kubernetes 1.29
- you want a scheduler that replaces kube-scheduler rather than complementing it

## Facets
- artifact type: service
- maturity: active
- function: scheduling, workflow-automation, monitoring, container-orchestration, middleware
- domain: cloud-computing, microservices, machine-learning, infrastructure-as-code
- platform: go, cloud, self-hosted
- tags: job-queueing, batch-scheduling, resource-quotas, gpu-scheduling, multi-tenancy, preemption, kubeflow, ray, fair-sharing, cncf, workload-admission, multi-cluster, containers, devops, kubernetes, docker

## Member repositories
- kubernetes-sigs/kueue (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:29.082461+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T07:34:27.520969+00:00, confidence not recorded.
  - readme: https://github.com/kubernetes-sigs/kueue (fetched 2026-08-28T04:07:29.082461+00:00, sha 83633f9f84c2)
  - homepage: https://kueue.sigs.k8s.io (fetched 2026-08-29T09:49:48.810518+00:00, sha 0a7c9ed7e0f9)
  - site_page: https://kueue.sigs.k8s.io/docs/getting-started/installation (fetched 2026-08-29T09:49:48.818488+00:00, sha 4c11e6d1eba3)
  - site_page: https://kueue.sigs.k8s.io/docs/getting-started/quick-start (fetched 2026-08-29T09:49:48.820663+00:00, sha 4ee93a07232f)
  - site_page: https://kueue.sigs.k8s.io/docs (fetched 2026-08-29T09:49:48.813048+00:00, sha b7a063cbd480)
  - site_page: https://kueue.sigs.k8s.io/v0.19/docs (fetched 2026-08-29T09:49:48.814903+00:00, sha 502a0d9b1e15)
  - site_page: https://kueue.sigs.k8s.io/v0.18/docs (fetched 2026-08-29T09:49:48.816762+00:00, sha 8a53242cb359)
- Data as of 2026-08-30T08:39:29.467469+00:00.