Dynamic Resource Allocation

Quota management and topology-aware placement for workloads using Kubernetes Dynamic Resource Allocation (DRA).

Dynamic Resource Allocation

Dynamic Resource Allocation (DRA) is a Kubernetes API for requesting and managing hardware devices such as GPUs, FPGAs, and network adapters. Kueue can account for DRA devices in quota management through two paths:

  1. ResourceClaimTemplate path: Pods explicitly reference a ResourceClaimTemplate that specifies a device request. Kueue maps each DeviceClass referenced by the claim to a logical resource name using deviceClassMappings in the Kueue Configuration.

  2. Extended resource path: Pods request DRA devices using the traditional resources.requests syntax (e.g., nvidia.com/gpu: 1). When the Kubernetes DeviceClass has an extendedResourceName field set (KEP-5004), the kube-scheduler automatically creates ResourceClaim objects from these requests. Kueue detects this and avoids double counting.

Which path should I use? If your workloads already use resources.requests for devices (e.g., nvidia.com/gpu: 1), use the extended resource path. If your workloads explicitly create ResourceClaimTemplate objects, use the ResourceClaimTemplate path.

How the ResourceClaimTemplate path works

Feature state beta since Kueue v0.18

When a Pod references a ResourceClaimTemplate, Kueue reads the deviceClassName from the template’s exactly field and looks it up in deviceClassMappings. With KueueDRAIntegrationPrioritizedList enabled it reads a request’s firstAvailable alternatives as well; see the limitations below for what that charges. This mapping tells Kueue which logical resource name to charge quota against. The number of units charged is determined by the count field in the device request (default 1).

Only the ExactCount allocation mode is supported. The All allocation mode is not supported.

For setup instructions, see Set Up Dynamic Resource Allocation.

How the extended resource path works

Feature state beta since Kueue v0.19

When a Pod requests an extended resource backed by DRA (e.g., nvidia.com/gpu: 1), the kube-scheduler auto-creates a ResourceClaim. Kueue detects the matching DeviceClass, uses extendedResourceName as the quota key, and drops the auto-created claim from accounting. This prevents quota from being charged for both the resources.requests entry and the auto-created claim, which would double count the same device. No deviceClassMappings configuration is needed; the mapping is discovered from the DeviceClass automatically. A deviceClassMappings entry covering that DeviceClass moves the charge to the mapping’s logical name.

This behavior is controlled by the KueueDRAIntegrationExtendedResource feature gate, which is enabled by default since v0.19.

Path separation

The two paths are independent:

  • ResourceClaimTemplate path: uses deviceClassMappings configuration.
  • Extended resource path: uses auto-discovery from DeviceClass objects.

Do not configure the same DeviceClass in both paths for the same workload. If overlap occurs, Kueue merges the resources using the deviceClassMappings logical name as the quota key, which may result in incorrect quota accounting.

Quota accounting

DRA resources are tracked in ClusterQueue quotas just like CPU or memory. The administrator includes the DRA resource name in coveredResources and sets a nominalQuota. Kueue supports three quota accounting modes:

  • Device count (default): Charges the count value from the device request (default 1 when omitted). A ClusterQueue with example.com/gpu: 8 allows up to 8 concurrent device allocations.
  • Counter-based: Charges the device’s consumesCounters value (e.g., GPU memory). See Counter-based quota.
  • Capacity-based: Charges the workload’s capacity.requests value rounded per the device’s RequestPolicy. See Capacity-based quota.

Admission and scheduling gap

There is a timing gap between Kueue admitting a workload (quota check) and the kube-scheduler allocating the actual device. Kueue does not know which specific device will be allocated — it only verifies that quota is available.

If the cluster state changes between these two steps (e.g., another system consumes the device), the scheduler may fail to allocate. The WaitForPodsReady feature provides a safety net by evicting workloads that fail to become ready within a configured timeout.

With Topology-Aware Scheduling, Kueue can also check before admission that a node has the devices a Pod needs. See Topology-Aware Scheduling with DRA.

Topology-Aware Scheduling with DRA

Feature state alpha since Kueue v0.20

KueueDRADeviceFeasibility adds a device check to Topology-Aware Scheduling (TAS): TAS places each Pod only on nodes that can allocate the devices it requests. Quota limits how many devices a ClusterQueue admits, not where they are, and without this feature TAS cannot tell which nodes have them. What goes wrong depends on how a Pod requests its devices:

  • ResourceClaimTemplate path: the devices are not in the Pod’s resource requests, so TAS places the Pod by its other resources alone. For example, a ClusterQueue with a quota of 8 GPUs on two nodes with 4 GPUs each admits a Pod whose ResourceClaimTemplate requests 6 GPUs: the quota allows it, but no node has 6. The Pod stays Pending while the workload holds the quota.
  • Extended resource path: the Pod requests a resource such as example.com/gpu, and TAS looks for it in each node’s allocatable, where a resource that only a DeviceClass provides never appears, so the workload fits on no node; see the warning in When the check runs.

How the device check works

  1. A workload is assigned a flavor with a topologyName (a TAS flavor).

  2. Because a Pod’s ResourceClaim objects do not exist before admission, Kueue builds the claims each Pod will need: from its ResourceClaimTemplates, or from the DeviceClass for an extended resource. For every node that the flavor and the Pod’s other scheduling constraints allow, Kueue tries to allocate these claims, using the same allocator as the kube-scheduler.

  3. Nodes where the allocation fails are dropped, and TAS places the Pods on the remaining nodes. For an extended resource, the check replaces TAS’s lookup in node allocatable, except on nodes that advertise the resource through a device plugin. This works for any topology, including one whose lowest level is not kubernetes.io/hostname.

  4. When no node is left, the workload stays pending, and its QuotaReserved condition message counts the nodes rejected for devices as draNoFit:

    couldn't assign flavors to pod set main: topology "dra-topology" doesn't allow to fit any of 1 pod(s). Total nodes: 2; excluded: draNoFit: 2
    
  5. Kueue checks the workload again when a ResourceSlice or DeviceClass changes, or when a ResourceClaim releases its devices.

When the check runs

WorkloadDevice check
Assigned a flavor with a topologyName, on the ResourceClaimTemplate or extended resource pathRuns
Assigned a flavor without a topologyNameDoes not run
In a ClusterQueue with a MultiKueue admission checkRuns on the worker cluster, where topology is assigned, not on the manager
In a ClusterQueue with a ProvisioningRequest admission checkSkipped on the first scheduling pass, which assigns no topology; runs on the second pass, after quota is reserved

Prerequisites

  • A Topology and a ResourceFlavor with topologyName, as described in Setup Topology-Aware Scheduling.
  • A DRA driver that publishes each node’s devices in ResourceSlice objects.
  • The KueueDRADeviceFeasibility feature gate enabled in Kueue Configuration. The gates it requires are enabled by default; if one of them is disabled, Kueue does not start and logs conflicting feature gates detected.

Device taints

Feature state alpha since Kueue v0.20

With this gate, the check skips devices with a NoSchedule or NoExecute device taint that the request does not tolerate, whether a DRA driver publishes the taint in a ResourceSlice or an administrator applies it with a DeviceTaintRule. None taints are ignored. A change to a DeviceTaintRule makes Kueue check rejected workloads again.

Kueue reads DeviceTaintRule objects only from Kubernetes 1.37 onwards, which serves them as resource.k8s.io/v1; on earlier versions it ignores taints from rules. With this gate disabled, it ignores all device taints. In both cases Kueue can admit a workload onto tainted devices that the kube-scheduler then refuses, so enable this gate together with KueueDRADeviceFeasibility.

Limitations of the check

  • One Pod per node: the check asks whether a node can serve one Pod of the PodSet, not how many. Kueue can place more Pods on a node than it has devices for, and the Pods that do not get a device stay Pending.
  • No release on preemption: devices held by workloads that Kueue would preempt are not freed in the check, so preemption cannot make a workload fit on devices.
  • Kubernetes DRA feature gates are read from the Kueue process: the check follows the gates of the Kubernetes version Kueue is built with, Kubernetes 1.37 for Kueue v0.20, not the cluster’s. This matters only when objects carry the fields of a DRA feature that the kube-scheduler has disabled, for example when the feature is disabled on the kube-scheduler but not on the kube-apiserver, or disabled after objects already used it. Otherwise the kube-apiserver drops the fields of a disabled feature, so Kueue and the kube-scheduler see the same devices.
  • Allocation time is not bounded: each check tries an allocation on every node. A slow DeviceClass CEL selector makes every scheduling cycle slower, rather than timing out.
  • Not every Kubernetes DRA feature is modeled: some, such as DRADeviceBindingConditions, change what the kube-scheduler does but not what the check predicts (full list).

For setup instructions, see Use Topology-Aware Scheduling with DRA.

MultiKueue

DRA workloads are supported with MultiKueue, except for firstAvailable requests; see the limitations below. MultiKueue syncs the workload and its owning job to worker clusters, but ResourceClaimTemplate and DeviceClass objects are not automatically synced. These must be created on each worker cluster separately by the cluster administrator.

Counter-based quota for partitionable devices

Feature state beta since Kueue v0.19

By default, Kueue tracks DRA quota by device count: each device request charges count units regardless of the device’s capacity. This means a small GPU partition and a full GPU both count as “1 device”, which does not reflect the actual resource consumption.

Kueue can track quota using counter values published by DRA drivers in ResourceSlice objects. This allows quota to reflect actual device capacity (e.g., GPU memory) rather than device count.

This behavior is controlled by the KueueDRAIntegrationPartitionableDevices feature gate, which is enabled by default since v0.19.

A DeviceClass uses either device-count quota (no sources configured) or counter-based quota (with sources), not both. Kueue rejects configurations that map the same DeviceClass to multiple resource names.

How it works

  1. The administrator configures a sources entry in deviceClassMappings that specifies which counter to track, which DRA driver to query, and a CEL expression to scope eligible devices.

  2. When a workload is submitted, Kueue reads the consumesCounters field from the matching devices in ResourceSlice objects to determine the actual counter charge.

  3. Kueue uses conservative charging: it takes the maximum consumesCounters value across all matched devices and multiplies by the request count. This ensures quota is not undercharged when different devices consume different amounts.

  4. The ClusterQueue quota is set in counter units (e.g., 800Gi for GPU memory) instead of device count.

Prerequisites

  • Kubernetes 1.35 or later with the DRAPartitionableDevices feature gate enabled (beta in Kubernetes 1.36).
  • A DRA driver that publishes consumesCounters on devices in ResourceSlice objects.

For setup instructions, see Set Up Dynamic Resource Allocation.

Counter-based vs capacity-based quota

Both modes track quota by actual resource consumption rather than device count, but they serve different device types:

Counter-based (PD)Capacity-based (CC)
Device typePartitioned devices (e.g., NVIDIA MIG)Shared devices (e.g., GPU time-slicing, MPS)
Charge sourceDevice’s consumesCountersWorkload’s capacity.requests
Who decides consumptionDriver (fixed per partition)User (variable per workload)
Upstream K8s featureKEP-4815 (DRAPartitionableDevices)KEP-5075 (DRAConsumableCapacity)

If your GPUs use hardware partitioning (MIG), use counter-based quota. If your GPUs allow software-level sharing where workloads request variable amounts of capacity, use capacity-based quota.

A cluster can use both modes simultaneously with different DeviceClasses using the same DRA driver. One DeviceClass with counter sources for partitioned devices and another with capacity sources for shared devices. Counter and capacity sources cannot be mixed within the same DeviceClass mapping.

Capacity-based quota for shared devices (consumable capacity)

Feature state alpha since Kueue v0.19

Some devices allow multiple workloads to share them simultaneously using software-level sharing mechanisms such as GPU time-slicing or MPS. These devices publish a Capacity field on each device in ResourceSlice objects (defined by KEP-5075) instead of using consumesCounters. Workloads specify how much capacity they need via capacity.requests on the device request.

Kueue can track quota using these capacity dimensions so that the total consumed capacity across all sharing workloads does not exceed the device’s published capacity.

This behavior is controlled by the KueueDRAIntegrationConsumableCapacity feature gate (Alpha, disabled by default in v0.19).

A DeviceClass uses either device-count quota (no sources), counter-based quota (with counter sources), or capacity-based quota (with capacity sources). Counter and capacity sources cannot be mixed in the same mapping.

How it works

  1. The administrator configures a capacity source entry in deviceClassMappings that specifies which capacity dimension to track, which DRA driver to query, and a CEL expression to scope eligible devices.

  2. When a workload is submitted, Kueue reads the workload’s capacity.requests from the ExactDeviceRequest for the configured dimension. If capacity.requests is omitted, Kueue uses the device’s RequestPolicy.Default or the full Capacity.Value as the charge.

  3. Kueue rounds the request per the device’s RequestPolicy (ValidValues or ValidRange with Step) to prevent quota gaming where a small request consumes more actual capacity after rounding by the kube-scheduler.

  4. For each matched device, Kueue computes the charge independently using the device’s own Default and policy, then takes the maximum across all devices. This ensures quota is never undercharged even if the deviceSelector matches heterogeneous devices.

  5. The ClusterQueue quota is set in capacity units (e.g., 800Gi for GPU memory) instead of device count.

Prerequisites

  • Kubernetes 1.36 or later with the DRAConsumableCapacity feature gate enabled (beta, enabled by default in Kubernetes 1.36).
  • A DRA driver that publishes Capacity and AllowMultipleAllocations on devices in ResourceSlice objects.
  • The KueueDRAIntegrationConsumableCapacity feature gate enabled in Kueue Configuration.

For setup instructions, see Set Up Dynamic Resource Allocation.

Limitations

The following limitations apply:

  • ResourceClaimTemplates only: Only ResourceClaimTemplate references are supported. Direct ResourceClaim references in the Pod spec are not supported and will result in inadmissible workloads.
  • ExactCount allocation mode only: the All allocation mode is not supported, in an exactly request or in an alternative of a firstAvailable one. Kueue reads a firstAvailable request only when the KueueDRAIntegrationPrioritizedList feature gate is enabled. That gate is alpha and off by default; see the note below for what it covers.
  • No device constraints or config: Device constraints (MatchAttribute) and per-request config are not supported.
  • No AdminAccess: Device requests with adminAccess: true are not supported.
  • TAS does not see devices: without Topology-Aware Scheduling with DRA, TAS may place a Pod on a node without the devices its ResourceClaimTemplate requests, and a workload that requests an extended resource only a DeviceClass provides fits on no node.
  • Device taints do not change quota: A tainted device is charged like any other. The per-node check can honor taints; see Device taints.
  • firstAvailable requests are charged, within limits: This support is experimental; do not enable it in production. With KueueDRAIntegrationPrioritizedList enabled, a firstAvailable request is charged once, the count every alternative asks for, which is what the scheduler allocates whichever alternative it picks. Every alternative of a request has to ask for the same count and map to the same logical resource; a request whose alternatives differ in count is refused, and so is an alternative on a mapping with a counter or capacity source. An alternative that sets capacity on the subrequest is charged its declared count like any other. Without KueueDRADeviceFeasibility, Kueue does not check that any alternative can be satisfied by the cluster, so a request whose alternatives are all infeasible holds its quota until the Workload is evicted, for example by WaitForPodsReady where it is configured; with it, such a Workload stays pending instead. The charge lands on the one ResourceFlavor the PodSet is assigned, and the Pods carry that flavor’s node labels, so keep every alternative’s devices behind the same flavors; with a flavor per device model, only the alternative with devices on the assigned flavor can run. MultiKueue does not support firstAvailable requests: a manager and a worker may resolve different templates, and nothing refuses such a Workload before dispatch yet.

Last modified September 30, 2026: Update main with the latest v0.20.0 (36c42c8df)