[Feature] [ML] GPU Jobs (#1549)

arangodb · Dec 21, 2023 · de4def4 · de4def4
1 parent df65970
commit de4def4
Show file tree

Hide file tree

Showing 5 changed files with 59 additions and 3 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -37,6 +37,7 @@
 - (Feature) (ML) Multi DB Settings
 - (Feature) (ML) Port adjustments
 - (Feature) (ML) Support for deployments with JWT auth enabled
+- (Feature) (ML) GPU Jobs
 
 ## [1.2.35](https://github.com/arangodb/kube-arangodb/tree/1.2.35) (2023-11-06)
 - (Maintenance) Update go-driver to v1.6.0, update IsNotFound() checks

diff --git a/docs/api/ArangoMLExtension.V1Alpha1.md b/docs/api/ArangoMLExtension.V1Alpha1.md
@@ -66,6 +66,16 @@ Links:
 
 ***
 
+### .spec.deployment.prediction.gpu
+
+Type: `boolean` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L31)</sup>
+
+GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+
+Default Value: `false`
+
+***
+
 ### .spec.deployment.prediction.image
 
 Type: `string` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/shared/v1/image.go#L31)</sup>
@@ -76,7 +86,7 @@ Image define image details
 
 ### .spec.deployment.prediction.port
 
-Type: `integer` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L30)</sup>
+Type: `integer` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L34)</sup>
 
 Port defines on which port the container will be listening for connections
 
@@ -122,6 +132,16 @@ Links:
 
 ***
 
+### .spec.deployment.project.gpu
+
+Type: `boolean` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L31)</sup>
+
+GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+
+Default Value: `false`
+
+***
+
 ### .spec.deployment.project.image
 
 Type: `string` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/shared/v1/image.go#L31)</sup>
@@ -132,7 +152,7 @@ Image define image details
 
 ### .spec.deployment.project.port
 
-Type: `integer` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L30)</sup>
+Type: `integer` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L34)</sup>
 
 Port defines on which port the container will be listening for connections
 
@@ -240,6 +260,16 @@ Links:
 
 ***
 
+### .spec.deployment.training.gpu
+
+Type: `boolean` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L31)</sup>
+
+GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+
+Default Value: `false`
+
+***
+
 ### .spec.deployment.training.image
 
 Type: `string` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/shared/v1/image.go#L31)</sup>
@@ -250,7 +280,7 @@ Image define image details
 
 ### .spec.deployment.training.port
 
-Type: `integer` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L30)</sup>
+Type: `integer` <sup>[\[ref\]](https://github.com/arangodb/kube-arangodb/blob/1.2.35/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go#L34)</sup>
 
 Port defines on which port the container will be listening for connections
 

diff --git a/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go b/pkg/apis/ml/v1alpha1/extension_spec_deployment_component.go
@@ -26,13 +26,24 @@ import (
 )
 
 type ArangoMLExtensionSpecDeploymentComponent struct {
+	// GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+	// +doc/default: false
+	GPU *bool `json:"gpu,omitempty"`
+
 	// Port defines on which port the container will be listening for connections
 	Port *int32 `json:"port,omitempty"`
 
 	// ContainerTemplate Keeps the information about Container configuration
 	*sharedApi.ContainerTemplate `json:",inline"`
 }
 
+func (s *ArangoMLExtensionSpecDeploymentComponent) GetGPU() bool {
+	if s == nil || s.GPU == nil {
+		return false
+	}
+	return *s.GPU
+}
+
 func (s *ArangoMLExtensionSpecDeploymentComponent) GetPort(def int32) int32 {
 	if s == nil || s.Port == nil {
 		return def

diff --git a/pkg/apis/ml/v1alpha1/zz_generated.deepcopy.go b/pkg/apis/ml/v1alpha1/zz_generated.deepcopy.go
diff --git a/pkg/crd/crds/ml-extension.schema.generated.yaml b/pkg/crd/crds/ml-extension.schema.generated.yaml
@@ -385,6 +385,9 @@ v1alpha1:
               prediction:
                 description: Prediction defines how Prediction workload will be deployed
                 properties:
+                  gpu:
+                    description: GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+                    type: boolean
                   image:
                     description: Image define image details
                     type: string
@@ -476,6 +479,9 @@ v1alpha1:
               project:
                 description: Project defines how Project workload will be deployed
                 properties:
+                  gpu:
+                    description: GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+                    type: boolean
                   image:
                     description: Image define image details
                     type: string
@@ -612,6 +618,9 @@ v1alpha1:
               training:
                 description: Training defines how Training workload will be deployed
                 properties:
+                  gpu:
+                    description: GPU defined if GPU Jobs are enabled for component. In use only for ArangoMLExtensionSpecDeploymentComponentPrediction and ArangoMLExtensionSpecDeploymentComponentTraining
+                    type: boolean
                   image:
                     description: Image define image details
                     type: string