DynamoComponentDeployment (DCD) Reference
Field reference for the DynamoComponentDeployment custom resource — the single Dynamo component that a DynamoGraphDeployment reconciles into pods.
A DynamoComponentDeployment (DCD) describes one component of a Dynamo inference graph — a frontend, a worker, a prefill or decode worker, a planner, or an EPP. The operator reconciles each DCD into a Kubernetes Deployment (or Grove workload) plus its Services, ConfigMaps, and pods.
A DCD is rarely authored on its own. Each entry in a DynamoGraphDeployment spec.components list is a DynamoComponentDeploymentSharedSpec, the same shape documented here under Shared component spec. The operator creates one child DCD per component. Author a standalone DCD only when you want to manage a single component’s lifecycle independently; otherwise define components inside a DGD.
This page documents the nvidia.com/v1beta1 API — the served, current version.
This reference covers user-configurable spec fields. For platform installation and operator configuration, see Install Dynamo.
Minimal example
Spec reference
The DCD spec is backendFramework plus the shared component spec inlined at the same level.
Inference backend framework for this component. Drives backend-specific defaults the operator injects into the spec.podTemplate.spec.containers[name=main] container.
Shared component spec
These fields are shared between a standalone DCD and each entry of a DGD spec.components list, except where a field is marked DGD-only. In a DGD, prefix them with spec.components[*]. Container paths below use the standalone DCD layout; replace the leading spec.podTemplate with spec.components[*].podTemplate for DGD components.
Customizes the primary Grove unit for a component embedded in a DGD. Standalone DCD OpenAPI omits this field. Use apiVersion: grove.io/v1alpha1; the target is PodCliqueTemplateSpec for a component backed by a PodClique or PodCliqueScalingGroupConfig for one backed by a scaling group. The value accepts topologyConstraint, the native component minimum (spec.minAvailable for standalone cliques or minAvailable for scaling groups), and rollingUpdate.maxUnavailable, separately or together. LPX conductor components support availability settings but not topology overrides. Omit target to let admission resolve it from the component shape. See Grove rollout budget overrides and ProviderOverride.
Stable logical identifier for the component, unique within its parent DGD’s spec.components list. Must match ^[A-Za-z0-9]([-A-Za-z0-9]*[A-Za-z0-9])?$, 1–63 characters. For a standalone DCD the defaulting webhook populates name from metadata.name, so you rarely set it explicitly. The name is decoupled from the underlying workload name so the operator can rename child workloads (for example, hash-suffixing worker DCDs during a rolling update) without losing the identity that labels, status maps, scaling adapters, planner RBAC, and EPP filters depend on.
Role of the component within the graph. Drives port mapping, frontend detection, planner RBAC, and the pod label nvidia.com/dynamo-component-type. prefill and decode are first-class values for disaggregated serving and can be set directly. At most one component per graph may be epp. Immutable after it is set. Standalone DCDs cannot use lpx.
The lpx type is experimental, requires the operator’s lpx.enabled setting, and may change incompatibly.
Experimental DGD-only compiled-model and scheduling configuration; requires the operator’s lpx.enabled setting and may change incompatibly. Set buildId; placement comes from the build manifest. Runtime options belong in role pod templates. See LPXConfig.
When the operator’s lpx.modelRegistryURL is configured, buildId must be relative to that registry; absolute paths and URLs require an empty registry URL.
Configures this LPX component’s scheduling attempts. Components may use different deadlines, including when they share a workload.
Best-effort deadline for each of this component’s LPU pipeline requests to leave Pending, including waiting for the scheduler to start. The timer uses a nonzero status.schedulingStartedAt when available; otherwise it uses metadata.creationTimestamp. A scheduler-provided timestamp for a later scheduling cycle takes precedence over the request’s original creation time. Valid values are 1 through 9,223,372,036; omission means unlimited. This field is not a solver budget or an overall deployment readiness timeout.
Current-generation Bound, NoFit, and Unsupported receipts are exempt. Degraded, Releasing, and Released receipts are also exempt when their committed execution was accepted for the current request generation. NoFit can therefore leave the deployment Pending indefinitely, and a Bound request can still wait indefinitely for runtime readiness.
Deadline handling runs after download checks, compiler-registry resolution, and desired-state validation. An outage or invalid desired configuration can delay failure reporting and cleanup indefinitely, even after the deadline has elapsed. The timer does not bound asynchronous cleanup or finalization.
Expiry records LPXSchedulingDeadlineExceeded before cleanup. Cleanup removes complete engine replicas only from a trailing suffix: an expired interior replica blocks deadline cleanup to preserve healthy higher ordinals. Explicit replicas authorizes lowering the Grove replica count to the surviving prefix; externally managed capacity remains unchanged. Grove scaling and owner garbage collection handle pod cleanup; the LPX graph controller does not delete pods directly. A later LPX input revision authorizes republication after cleanup. An existing failure remains cleanup authority for cycles that started before it, so retiring those cycles does not consume the retry edit.
Groups opt-in LPX options whose shape may change incompatibly between v1beta1 releases. These fields are not covered by the v1beta1 deprecation policy.
Selects partitions of a hybrid build that the Cyborg conductor runs on its own GPU. Set mode: All to run every partition on the GPU, or mode: IDs with ids to run the listed partitions. ids is required with IDs and forbidden with All. Omitting localPartitions runs every partition on LPUs. Hybrid XT (LP20) and HX (LP30) builds support this field. LPU-only and speculative builds reject it.
The operator schedules Agent Pods and LPU pipeline requests only for the remaining partitions, and lists only those partitions in the Agents’ partition configuration and, for XT builds, the conductor’s lpu_servers. For HX builds, the request’s allocation metadata also lists only the remaining partitions. When every partition is local, it renders no Agent PodClique and creates no LPU pipeline request. It publishes the resolved local partitions to the Cyborg main container as LPX_LOCAL_PARTITION_IDS, a comma-separated list of compiler partition IDs.
ids lists compiler partition IDs of the build’s runtime partitions. A selected prop-sync chain is one runtime partition: select its first partition, which keeps the whole chain on the GPU. Unknown IDs and non-first chain members are rejected during reconciliation. Because a chain moves as a unit, a local selection never splits a chain. The operator keeps a prop-sync connector only when both of its partitions remain on LPUs, and renumbers it to their positions among the remaining partitions. Changing the selection changes the workload digest and rolls the workload.
Declares the Dynamo runtime version in spec.podTemplate.spec.containers[name=main].image by default, or spec.podTemplate.spec.initContainers[name=runtime].image when the dynamo sidecar is present, taking precedence over the image tag. Use canonical MAJOR.MINOR.PATCH, for example "1.4.0", with each part between 0 and 9999 and no leading zeros, v prefix, prerelease suffix, or build metadata. This field does not change the image. See Runtime Version Compatibility for admission requirements and rollout behavior.
Desired number of complete component instances. For a single-node component, each instance is one pod. For a multinode component, each instance contains all of its roles. Minimum 0. When scalingAdapter is set, this field is owned by the DynamoGraphDeploymentScalingAdapter and should not be modified directly.
For a single type: lpx component in a DGD, this counts complete engine replicas, not individual role Pods. Hybrid and LPU-only engines allow up to 2496 replicas; generated Pod hostname limits may lower that maximum. In a speculative pair, draft replicas count model instances (1–8), and target replicas must be 1. LPX does not support scaling adapters or scale-to-zero.
Minimum complete component replicas guaranteed to be gang-scheduled before a shortfall triggers gang termination. Deprecated: use the native minimum through the component providerOverride. Supported only for Grove-backed DGD components; rejected for non-Grove deployments. Existing legacy deployments retain this field and its default of 1. New Grove DGDs without any legacy minimum resolve an omitted native minimum to 1 during rendering, without creating a provider override. Minimum 1; the effective value is immutable after creation, including when migrating to the native form. Positive replicas must be at least this value. Scaling to 0 is allowed and retains minAvailable for the next scale-up.
Under Coherent updates, this value also defines the component’s minimum viable replacement unit. With no surge capacity, replacing that unit can make this many replicas unavailable together. Both minimum forms use RollingRecreate unless the DGD strategy annotation explicitly selects Coherent. Migrating component minima to provider overrides leaves the update strategy unchanged. See Configure native minimum availability and Coherent capacity and disruption.
Under Coherent updates, this value also defines the component’s minimum viable replacement unit. With no surge capacity, replacing that unit can make this many replicas unavailable together. Start with 1 for new deployments unless a larger viable unit is required. See Coherent capacity and disruption.
For LPX components with omitted replicas, minAvailable sets the initial Grove scaling-group size; subsequent capacity is externally managed.
Complete Pod template shared by all roles of the component. It is mutually exclusive with roles[*].podTemplate. Every component must include spec.podTemplate.spec.containers[name=main] with a non-empty image. By default the operator injects command, environment, port, probe, resource, and volume-mount defaults into spec.podTemplate.spec.containers[name=main], merging your overrides by name. Declaring spec.podTemplate.spec.initContainers[name=runtime] activates Dynamo Sidecar Mode. Existing standard-mode components created without spec.podTemplate.spec.containers[name=main].image may retain that omission on unrelated updates. For DGD components whose runtime image has no parseable semantic-version tag, set runtimeVersionOverride. Other containers are user-managed and must specify their required fields, including image; frontendSidecar can explicitly select a regular container for frontend defaults. Replaces the ten separate per-component fields (resources, envs, livenessProbe, and so on) that existed in v1alpha1.
For type: lpx, omit this field and use roles[].podTemplate. Every LPX role requires its own template with an explicit main container image.
For runtime 1.6.0 and later, frontend containers receive a startup probe that allows five minutes for NATS and etcd connections and initialization. The same default applies to a designated frontend sidecar, using its own image version independently of the parent component’s runtime version and runtimeVersionOverride. Older or unknown runtime versions do not receive this default.
Each supplied startupProbe, livenessProbe, or readinessProbe is a complete value that replaces the applicable default. It must include exactly one handler (exec, httpGet, tcpSocket, or grpc). In both v1alpha1 and v1beta1, the DGD and DCD APIs reject probes that only change timing or define multiple handlers, including probes in role templates and sidecars. To change timing, include the handler and the desired timing settings. Omit a probe to retain the applicable operator default. See Health Probes for probe settings and timeout adjustments.
Configures a worker, prefill, or decode component that spans multiple pods. Other component types reject this field. A pre-existing unsupported combination can remain unchanged during unrelated updates or remove multinode, but cannot change or reintroduce it. See Multinode Orchestration.
Number of nodes to deploy. Minimum 2 and immutable after creation. Total GPUs used is nodeCount × container GPU request.
Named Pod-producing parts of a compound component, keyed by name. Multinode components must contain exactly one leader and one worker; their cardinality is fixed by multinode.nodeCount, and omitted role counts default to 1 and multinode.nodeCount - 1, respectively. Omit the field to retain the implicit leader/worker layout. If either multinode role sets podTemplate, both roles must set one and the component-level podTemplate must be absent.
LPX components use agent for Agent Pods and conductor for the serving runtime. Independent LPX components each declare a conductor role with an explicit template. In a speculative pair, the draft has only an agent role and the target declares both roles. A hybrid engine’s GPU runtime uses the conductor template.
Standalone DCD roles cannot use LPX templates or DGD-only provider overrides. See ComponentRoleSpec.
Semantic role name: leader or worker for ordinary multinode components, conductor or agent for LPX components. Names must be unique within the component.
Logical cardinality of this role in one complete component instance; minimum 1. For ordinary multinode components, admission defaults and persists an omitted value as leader: 1 or worker: multinode.nodeCount - 1; an explicit value must match that fixed shape. For LPX, omitted conductor replicas use 1, subject to runtime-specific limits. Agent replicas are derived from the model build and lpx.experimental.localPartitions; an explicit count must match that derived count.
Complete Pod template for this role. It must include a container named main with a non-empty image. Role templates do not inherit from each other or from the component-level template.
An init container named runtime is rejected in role templates because Dynamo sidecar mode does not yet support multinode or LPX components.
For multinode components, supplying complete role templates transfers ownership of backend-specific leader and worker commands and topology arguments to the user; Dynamo skips its vLLM, SGLang, and TensorRT-LLM role-dependent launch rewrites. Common Pod and workload wiring remains operator-owned. For vLLM multiprocessing, Dynamo still injects the shared coordination port 29500. Multinode role templates cannot currently be combined with GMS or failover. The component remains the lifecycle, rollout, scaling, service, and status boundary.
For LPX components, every declared role requires its own template. Agent and conductor templates are independent.
core/v1.PodTemplateSpecProvider configuration for the workload unit generated for this role. Available only on ordinary multinode components embedded in a DGD. LPX roles do not support it; standalone DCD OpenAPI omits this field. For Grove multinode leader and worker roles, use target PodCliqueTemplateSpec; only topologyConstraint is allowed in value. Role overrides are not supported for inter-pod GMS components. See ProviderOverride.
Size of the tmpfs mounted at /dev/shm. Omit to use the operator default (8Gi); set a positive quantity for a custom size; set "0" to disable the shared-memory volume entirely.
Places the component in the global Dynamo namespace (dynamo) rather than the per-deployment namespace. By default, the operator combines the Kubernetes namespace and DGD name with a hyphen and replaces dots in the DGD name with hyphens. If the result exceeds 63 characters, the operator shortens it and appends a 16-character hash derived from the original namespace/name pair. Values of 63 characters or fewer remain unchanged.
References a model served by this component. When set, a headless service is created for endpoint discovery.
Base model identifier, for example llama-3-70b-instruct-v1.
Model revision or version.
Opts the component into a DynamoGraphDeploymentScalingAdapter (DGDSA). Set it — even as an empty object, scalingAdapter: {} — to create a DGDSA that owns replicas so external autoscalers (HPA, KEDA, Planner) can drive scaling through the Scale subresource. Omit the field to opt out.
Designates a container in each selected complete Pod template as the frontend sidecar. The value must match a container name in the component template or every role template; the operator merges its frontend-sidecar defaults (Dynamo env vars, ports, health probes) into that container the same way it merges into main.
Strict namespace-prefix matching uses the sidecar’s own image version, independently of the component’s runtime container and runtimeVersionOverride. If the sidecar image tag does not identify the runtime version, set DYN_NAMESPACE_PREFIX_STRICT=true in that container’s env when its image contains the strict-prefix implementation.
Configures a PVC-backed compilation cache. The operator handles backend-specific mount paths and environment variables, so you do not hand-wire them into podTemplate.
Name of a user-created PVC, which must exist in the same namespace as the deployment.
Overrides the backend-specific default mount path. When empty, the operator selects a default appropriate for the backend framework.
Cached artifacts (for example vLLM’s torch.compile/CUDA graph cache) are specific to the
backend, torch, and CUDA versions that produced them. After upgrading to a new Dynamo release —
or any change to the underlying backend image — clear the PVC’s contents before redeploying.
Stale entries from an older build have caused worker crashes at startup (for example an NVML
assertion failure during CUDA graph capture) that disappear once the cache is emptied.
Deprecated: omit eppConfig and use the native Rust EPP, which is configured through environment variables and takes no config file. Presence of this field selects the legacy Go EPP pod contract, so existing deployments keep running across an operator upgrade until you clear it.
Only valid when type is epp, and its use is decided by the component’s resolved runtime version: required below 1.5.0 (legacy Go EPP image), forbidden at 1.5.0 and later (native Rust EPP image). Migrate by clearing eppConfig and moving to a 1.5.0+ image in the same update. Exactly one of configMapRef or config must be set. See the Gateway API Routing Reference.
References a user-provided ConfigMap key containing EPP configuration. Mutually exclusive with config.
EPP EndpointPickerConfig supplied inline. The operator marshals it to YAML and creates the ConfigMap for you. Mutually exclusive with configMapRef.
Component-level topology placement. See the SpecTopologyConstraint API.
Topology domain to pack pods within. Must match a domain defined in the referenced ClusterTopology. When the parent DGD also sets spec.topologyConstraint.packDomain, this value must be narrower than or equal to it.
Opt-in preview features whose API shape may change in breaking ways between v1beta1 releases. Fields here are not covered by the normal v1beta1 deprecation policy — do not rely on them for production workloads.
See: ExperimentalSpec
Dynamo Sidecar Mode
Declaring spec.podTemplate.spec.initContainers[name=runtime] activates Dynamo sidecar mode. You must declare this init container yourself; the operator merges defaults into it and does not create it. It must specify a non-empty image and restartPolicy: Always. The name alone selects the mode; a missing restart policy is rejected. spec.podTemplate.spec.containers[name=runtime] or an init container with another name does not activate it.
The operator injects Dynamo environment variables, identity, transport TLS, system port, and startup, liveness, and readiness probes into spec.podTemplate.spec.initContainers[name=runtime]. The spec.podTemplate.spec.containers[name=main] container runs the engine with user-provided launch configuration and probes. Graph-level environment variables apply to both containers; compilation cache and shared-memory configuration remain on spec.podTemplate.spec.containers[name=main]. A separate frontendSidecar is supported.
Both containers use the existing environment merge behavior: merged entries are sorted by name and de-duplicated, with container-level values overriding graph-level values and runtime defaults. The operator origin version does not change this behavior. Backend-specific additions retain their existing placement; for example, VLLM_CACHE_ROOT is appended to the engine environment after merging.
This mode supports worker, prefill, and decode components. Multinode, enabled checkpoint, GPU memory service, and failover are rejected because they are not currently supported in this mode. Support for these features is planned for a future release. Other component types cannot declare spec.podTemplate.spec.initContainers[name=runtime].
The same convention applies to v1alpha1 through spec.extraPodSpec.initContainers[name=runtime]. Conversion preserves the live init-container list. Adding, renaming, or removing spec.podTemplate.spec.initContainers[name=runtime] changes the mode and can trigger a worker rollout. Reserve this name for the Dynamo runtime; rename unrelated init containers before upgrading the operator.
Set the engine launch configuration and sidecar command for your backend. See the vLLM sidecar manifests for complete examples.
Runtime Version Compatibility
DGD admission requires runtimeVersionOverride for non-LPX components when the runtime image has no parseable semantic-version tag. The runtime image is spec.components[*].podTemplate.spec.initContainers[name=runtime].image in Dynamo sidecar mode, or spec.components[*].podTemplate.spec.containers[name=main].image otherwise. Also set it when a parseable tag does not represent the Dynamo runtime version in the image. The override must describe the Dynamo runtime actually contained in the image.
When updating an image to a different Dynamo runtime version, update any configured runtimeVersionOverride in the same change, or remove it if the new image tag correctly identifies the runtime version. The operator uses the resolved runtime version to select feature gates that control how it renders the component’s PodSpec, including flags, environment variables, and probes. For Dynamo 1.5.0 and later, changing the override alone may change the rendered PodSpec or worker revision and trigger a rollout, even when the image reference is unchanged. Keep the override aligned with the Dynamo runtime actually contained in the image.
For components generated by a DGDR, an explicit component override takes precedence over the DGDR-level default. See Profiler image version compatibility.
Status
The operator maintains observed state under status.
Standard Kubernetes conditions. Available reports whether the component is serving traffic; DynamoComponentReady reports whether the underlying Dynamo component is ready.
Most recent metadata.generation the controller has reconciled. A component is up to date when this equals metadata.generation.
Replica status for this component: desired, ready, and available counts. Shares the shape documented on the DGD Reference.
Additional Types
Nested struct types referenced by the fields above, broken out here to keep the field lists shallow. Types prefixed core/v1., metav1., resource., or runtime. are standard Kubernetes types and link to their Go package documentation instead of being expanded.
ExperimentalSpec
Groups opt-in preview features for a component. Referenced by experimental. Nested types (GMSClientPodSpec, ComponentCheckpointJobConfig, DynamoCheckpointIdentity) are expanded inline under the field that references them.
These preview features can change or disappear between v1beta1 releases without a name-preserving graduation path. They are excluded from the v1beta1 deprecation policy — do not rely on them for production workloads.
Configures the GPU Memory Service (GMS). When set, GPU access for GMS clients is managed through Dynamic Resource Allocation (DRA), and the operator replaces the spec.podTemplate.spec.containers[name=main] container’s GPU resources with a DRA ResourceClaim.
Selects the GMS deployment topology.
Allowed values: IntraPod InterPodThe DRA DeviceClass to request GPUs from.
Additional user-declared containers that should be wired as GMS clients in service pods. SnapshotJob capture Pod clients are declared under checkpoint.job.gmsClientContainers instead. In each rendered pod, only matching container names are wired; absent names are ignored. Each name must match ^[a-z0-9]([-a-z0-9]*[a-z0-9])?$, 1–63 characters.
Additional GMS client pods for inter-pod GMS. Reserved for future use and rejected until inter-pod client orchestration is wired.
Identifies this client pod. Must match ^[a-z0-9]([-a-z0-9]*[a-z0-9])?$, 1–63 characters.
Configures the pod to run as a GMS client.
core/v1.PodTemplateSpecConfigures active-passive GPU failover for a worker component. The spec.podTemplate.spec.containers[name=main] container is cloned into two engine containers (active + standby) sharing GPUs via DRA, and the standby acquires the flock when the active engine fails. Requires gpuMemoryService to be set, failover.mode to match gpuMemoryService.mode, and the nvidia.com/dynamo-kube-discovery-mode: container annotation on the DGD.
Failover deployment topology. Must match the GMS mode on the same component.
Number of shadow (standby) engine containers per rank. Reserved for future use; the operator currently creates exactly one shadow. Minimum 1, maximum 1.
Configures experimental Grove workload rendering for a component in a DGD. Requires the Grove pathway.
When true, renders a single-node component as a PodCliqueScalingGroup with one single-pod PodClique per replica. The first minAvailable replicas join the deployment’s base PodGang; each additional replica gets a separate PodGang. false or omission retains automatic selection: multinode and inter-pod GMS components use scaling groups, while other single-node components use a PodClique. Immutable after creation.
Experimental process checkpoint/restore through standalone Snapshot. The API retains the name checkpoint; configure it under spec.components[*].experimental in a DGD. Enable checkpoint support in the Dynamo operator and install Snapshot before using this field.
Set enabled: true without checkpointRef for DGD-managed capture through a SnapshotJob. Set checkpointRef to restore an existing compatible PodSnapshot in the same namespace. Omit the deprecated mode and identity fields.
Dynamo v1.5.0 is a hard compatibility boundary for checkpoint resources. Dynamo creates automatic captures through the standalone Snapshot operator’s SnapshotJob API, and checkpointRef names a standalone PodSnapshot. Legacy DynamoCheckpoint objects and their artifacts cannot be restored.
The upgrade does not delete the legacy DynamoCheckpoint CRD, instances, PodSnapshot objects, PodSnapshotContent objects, PVC data, or stored artifacts. Before upgrading, list retained DynamoCheckpoint objects and legacy PodSnapshot objects labeled nvidia.com/snapshot-owner, recreate any required snapshots through the standalone Snapshot APIs, and delete unneeded legacy checkpoints while the old controller can still run their finalizers. After upgrading, review and remove unsupported resources manually. Do not delete the shared PodSnapshot or PodSnapshotContent CRDs because the standalone Snapshot operator uses them.
Whether checkpointing is enabled for this component. When true, omit checkpointRef for a DGD-managed automatic checkpoint, or set checkpointRef to restore a PodSnapshot in the same namespace. Omit the checkpoint block, or set enabled: false, to disable checkpointing.
When serving workers start relative to automatic capture. Immediate starts them cold while the SnapshotJob runs; only Pods created after capture completes restore. Snapshot readiness does not restart existing workers. WaitForCheckpoint keeps serving replicas at zero until capture completes and the PodSnapshot is Ready, then starts them from it. Capture completion includes any helper containers that persist state.
Whether DGD-managed capture resources and snapshot artifacts are deleted or retained when the owning DGD is deleted. Explicit checkpointRef PodSnapshots are never owned or deleted by the DGD, and retained automatic checkpoints are not valid checkpointRef targets.
Names an existing PodSnapshot in the same namespace. Before workers can restore, the snapshot must be Ready and carry matching Dynamo compatibility metadata for the target worker configuration. If the named snapshot does not exist, its capture has failed, or its compatibility metadata does not match, the operator reports an error and does not restore workers from it. Both startup policies perform these checks.
Standalone worker-class (worker, prefill, or decode) DynamoComponentDeployment resources cannot set this field; configure it on the component in the owning DGD. Do not set job with checkpointRef.
The workload container to snapshot and restore. Must match ^[a-z0-9]([-a-z0-9]*[a-z0-9])?$, 1–63 characters.
Automatic captures run this container under Snapshot’s cuInterpose launcher, so it must set command. The main container always has one, because the operator defaults it to /bin/sh -c. To opt out, set the pod template annotation nvidia.com/cuda-shared-memory-support: disabled. Multi-GPU captures can then fail when NCCL uses cuMem or NVLS memory.
Customizes the DGD-managed SnapshotJob capture Pod.
SnapshotJob capture Pod containers that should receive GMS client wiring. Requires gpuMemoryService on the component. Each name must match ^[a-z0-9]([-a-z0-9]*[a-z0-9])?$, 1–63 characters.
Customizes the SnapshotJob capture Pod. The operator starts from the selected workload container and merges this template, so you can add helper containers such as gms-saver.
Deprecated: omit mode. Use enabled: true without checkpointRef for a DGD-managed automatic checkpoint, or use checkpointRef to restore a named PodSnapshot.
Deprecated: omit for DGD-managed checkpoints; the operator ignores this field. Use checkpointRef to restore an existing PodSnapshot.
Model identifier, for example meta-llama/Llama-3-70B.
Runtime framework.
Allowed values: vllm sglang trtllmDynamo platform version.
Tensor parallel configuration. Deprecated: checkpoint launch uses the pod template instead. Minimum 1.
Pipeline parallel configuration. Deprecated: checkpoint launch uses the pod template instead. Minimum 1.
Data type, for example fp16, bf16, or fp8.
Maximum sequence length. Minimum 1.
Legacy identity parameters. Ignored for DGD-managed automatic capture.