Skip to content

fix: increase shm default size and make it configurable #2616

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
julienmancuso merged 3 commits into from
Aug 24, 2025
Merged

fix: increase shm default size and make it configurable #2616

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
74aaf23
fix: increase shm default size and make it configurable
julienmancuso Aug 21, 2025
36c40ba
fix: increase shm default size and make it configurable
julienmancuso Aug 21, 2025
d7d49a9
fix: increase shm default size and make it configurable
julienmancuso Aug 21, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
fix: increase shm default size and make it configurable
  • Loading branch information
@julienmancuso
julienmancuso committed Aug 21, 2025
commit 74aaf239466419e2cdf5ea79c86ea2844814f324
12 changes: 12 additions & 0 deletions deploy/cloud/helm/crds/templates/nvidia.com_dynamocomponentdeployments.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10241,6 +10241,18 @@ spec:
serviceName:
description: contains the name of the component
type: string
sharedMemory:
description: SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size).
properties:
disabled:
type: boolean
size:
anyOf:
- type: integer
- type: string
pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
x-kubernetes-int-or-string: true
type: object
type: object
status:
description: Status reflects the current observed state of the component deployment.
Expand Down
12 changes: 12 additions & 0 deletions deploy/cloud/helm/crds/templates/nvidia.com_dynamographdeployments.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10340,6 +10340,18 @@ spec:
serviceName:
description: contains the name of the component
type: string
sharedMemory:
description: SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size).
properties:
disabled:
type: boolean
size:
anyOf:
- type: integer
- type: string
pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
x-kubernetes-int-or-string: true
type: object
type: object
description: |-
Services allows per-service overrides of the component deployment settings.
Expand Down
5 changes: 5 additions & 0 deletions deploy/cloud/operator/api/v1alpha1/common.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,3 +44,8 @@ type Autoscaling struct {
Behavior *autoscalingv2.HorizontalPodAutoscalerBehavior `json:"behavior,omitempty"`
Metrics []autoscalingv2.MetricSpec `json:"metrics,omitempty"`
}

type SharedMemorySpec struct {
Disabled bool `json:"disabled,omitempty"`
Size resource.Quantity `json:"size,omitempty"`
}
3 changes: 3 additions & 0 deletions deploy/cloud/operator/api/v1alpha1/dynamocomponentdeployment_types.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -92,6 +92,9 @@ type DynamoComponentDeploymentSharedSpec struct {
// Ingress config to expose the component outside the cluster (or through a service mesh).
Ingress *IngressSpec `json:"ingress,omitempty"`

// SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size).
SharedMemory *SharedMemorySpec `json:"sharedMemory,omitempty"`

// +optional
// ExtraPodMetadata adds labels/annotations to the created Pods.
ExtraPodMetadata *dynamoCommon.ExtraPodMetadata `json:"extraPodMetadata,omitempty"`
Expand Down
21 changes: 21 additions & 0 deletions deploy/cloud/operator/api/v1alpha1/zz_generated.deepcopy.go
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

12 changes: 12 additions & 0 deletions deploy/cloud/operator/config/crd/bases/nvidia.com_dynamocomponentdeployments.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10241,6 +10241,18 @@ spec:
serviceName:
description: contains the name of the component
type: string
sharedMemory:
description: SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size).
properties:
disabled:
type: boolean
size:
anyOf:
- type: integer
- type: string
pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
x-kubernetes-int-or-string: true
type: object
type: object
status:
description: Status reflects the current observed state of the component deployment.
Expand Down
12 changes: 12 additions & 0 deletions deploy/cloud/operator/config/crd/bases/nvidia.com_dynamographdeployments.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10340,6 +10340,18 @@ spec:
serviceName:
description: contains the name of the component
type: string
sharedMemory:
description: SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size).
properties:
disabled:
type: boolean
size:
anyOf:
- type: integer
- type: string
pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
x-kubernetes-int-or-string: true
type: object
type: object
description: |-
Services allows per-service overrides of the component deployment settings.
Expand Down
4 changes: 2 additions & 2 deletions deploy/cloud/operator/internal/controller/dynamocomponentdeployment_controller_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -793,7 +793,7 @@ func TestDynamoComponentDeploymentReconciler_generateLeaderWorkerSet(t *testing.
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(5*1024*1024*1024, resource.BinarySI), // 5gi (calculated from memory limit / 4)
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -893,7 +893,7 @@ func TestDynamoComponentDeploymentReconciler_generateLeaderWorkerSet(t *testing.
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(5*1024*1024*1024, resource.BinarySI), // 5gi (calculated from memory limit / 4)
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down
36 changes: 15 additions & 21 deletions deploy/cloud/operator/internal/dynamo/graph.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -778,9 +778,10 @@ func GenerateBasePodSpec(
MountPath: *component.PVC.MountPoint,
})
}
shmVolume, shmVolumeMount := generateSharedMemoryVolumeAndMount(&container.Resources)
volumes = append(volumes, shmVolume)
container.VolumeMounts = append(container.VolumeMounts, shmVolumeMount)
if shmVol, shmMount := generateSharedMemoryVolumeAndMount(component.SharedMemory); shmVol != nil && shmMount != nil {
volumes = append(volumes, *shmVol)
container.VolumeMounts = append(container.VolumeMounts, *shmMount)
}

// Apply backend-specific container modifications
multinodeDeployer := MultinodeDeployerFactory(multinodeDeploymentType)
Expand Down Expand Up @@ -1179,36 +1180,29 @@ func GenerateBasePodSpecForController(
return podSpec, nil
}

func generateSharedMemoryVolumeAndMount(resources *corev1.ResourceRequirements) (corev1.Volume, corev1.VolumeMount) {
sharedMemorySizeLimit := resource.MustParse("512Mi")
// Check if we have memory limits to work with
memoryLimit := resources.Limits[corev1.ResourceMemory]
if !memoryLimit.IsZero() {
// Use 1/4 of memory limit
calculatedSize := resource.NewQuantity(memoryLimit.Value()/4, resource.BinarySI)
// Apply bounds: minimum 512Mi, maximum 8Gi
minSize := resource.MustParse("512Mi")
maxSize := resource.MustParse("8Gi")

if calculatedSize.Cmp(minSize) > 0 && calculatedSize.Cmp(maxSize) < 0 {
sharedMemorySizeLimit = *calculatedSize
} else if calculatedSize.Cmp(maxSize) >= 0 {
sharedMemorySizeLimit = maxSize // Cap at maximum
func generateSharedMemoryVolumeAndMount(spec *v1alpha1.SharedMemorySpec) (*corev1.Volume, *corev1.VolumeMount) {
// default: enabled=true, size=8Gi
size := resource.MustParse("8Gi")
if spec != nil {
if spec.Disabled {
return nil, nil
}
if !spec.Size.IsZero() {
size = spec.Size
}
// If calculatedSize < minSize, keep the 512Mi base
}
volume := corev1.Volume{
Name: commonconsts.KubeValueNameSharedMemory,
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: &sharedMemorySizeLimit,
SizeLimit: &size,
},
},
}
volumeMount := corev1.VolumeMount{
Name: commonconsts.KubeValueNameSharedMemory,
MountPath: "/dev/shm",
}
return volume, volumeMount
return &volume, &volumeMount
}
20 changes: 10 additions & 10 deletions deploy/cloud/operator/internal/dynamo/graph_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1235,7 +1235,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(536870912, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -1378,7 +1378,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(536870912, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -1733,7 +1733,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -1883,7 +1883,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -1989,7 +1989,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -2134,7 +2134,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -2509,7 +2509,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -2648,7 +2648,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -2755,7 +2755,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down Expand Up @@ -2899,7 +2899,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{
Medium: corev1.StorageMediumMemory,
SizeLimit: resource.NewQuantity(512*1024*1024, resource.BinarySI),
SizeLimit: resource.NewQuantity(8*1024*1024*1024, resource.BinarySI),
},
},
},
Expand Down
Loading