Skip to content

feat(nvca): add storage-aware model cache runtime #1357

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

Open
balajinvda wants to merge 3 commits into
base: feat/nvca-storage-capability-catalog
Choose a base branch
from
Open

feat(nvca): add storage-aware model cache runtime #1357

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0824274
feat(nvca): persist durable model cache storage decisions
balaji-g Aug 30, 2026
559a81b
feat(nvca): add RWX read-only model cache runtime
balaji-g Aug 30, 2026
0917a4d
docs(nvca): clarify shared writer metadata
balaji-g Aug 30, 2026
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
32 changes: 26 additions & 6 deletions deploy/helm/nvca-operator/nvca-operator/files/nvcf-storage-capabilities-v1alpha1.schema.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,15 +16,19 @@
"properties": {
"nvmesh-csi.excelero.com": {"$ref": "#/$defs/driver"}
},
"additionalProperties": {"$ref": "#/$defs/disabledDriver"}
"additionalProperties": {"$ref": "#/$defs/nonNVMeshDriver"}
}
},
"$defs": {
"accessMode": {
"type": "string",
"enum": ["ReadWriteOnce", "ReadOnlyMany", "ReadWriteMany"]
},
"transitionStrategy": {
"regularTransitionStrategy": {
"type": "string",
"enum": ["disabled", "nvmesh", "rwxReadOnly"]
},
"helmTransitionStrategy": {
"type": "string",
"enum": ["disabled", "nvmesh"]
},
Expand All @@ -33,8 +37,8 @@
"additionalProperties": false,
"required": ["regularModelCache", "helmModelCache"],
"properties": {
"regularModelCache": {"$ref": "#/$defs/transitionStrategy"},
"helmModelCache": {"$ref": "#/$defs/transitionStrategy"}
"regularModelCache": {"$ref": "#/$defs/regularTransitionStrategy"},
"helmModelCache": {"$ref": "#/$defs/helmTransitionStrategy"}
}
},
"driver": {
Expand Down Expand Up @@ -79,17 +83,33 @@
}
}
}
},
{
"if": {
"properties": {
"transitions": {
"properties": {"regularModelCache": {"const": "rwxReadOnly"}},
"required": ["regularModelCache"]
}
},
"required": ["transitions"]
},
"then": {
"properties": {
"accessModes": {"contains": {"const": "ReadWriteMany"}}
}
}
}
]
},
"disabledDriver": {
"nonNVMeshDriver": {
"allOf": [
{"$ref": "#/$defs/driver"},
{
"properties": {
"transitions": {
"properties": {
"regularModelCache": {"const": "disabled"},
"regularModelCache": {"enum": ["disabled", "rwxReadOnly"]},
"helmModelCache": {"const": "disabled"}
}
}
Expand Down
3 changes: 2 additions & 1 deletion deploy/helm/nvca-operator/nvca-operator/files/nvcf-storage-capabilities-v1alpha1.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,8 @@
# This public catalog is owned by NVCA and installed with the NVCA chart.
# accessModes lists only modes demonstrated for the exact tested provider.
# A disabled transition means that the end-to-end workflow is not qualified.
# The catalog is not yet wired into NVCA reconciliation or backend selection.
# NVCA reads this catalog when it creates a new model-cache storage selection.
# Only a registered, non-disabled transition can select durable caching.
apiVersion: storage.nvcf.nvidia.com/v1alpha1
kind: StorageCapabilityCatalog
drivers:
Expand Down
818 changes: 455 additions & 363 deletions docs/dev/sdd-storage-agnostic-cache-architecture.md
View file
Open in desktop

Large diffs are not rendered by default.

32 changes: 26 additions & 6 deletions ...vices/nvca/deployments/nvca-operator/files/nvcf-storage-capabilities-v1alpha1.schema.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,15 +16,19 @@
"properties": {
"nvmesh-csi.excelero.com": {"$ref": "#/$defs/driver"}
},
"additionalProperties": {"$ref": "#/$defs/disabledDriver"}
"additionalProperties": {"$ref": "#/$defs/nonNVMeshDriver"}
}
},
"$defs": {
"accessMode": {
"type": "string",
"enum": ["ReadWriteOnce", "ReadOnlyMany", "ReadWriteMany"]
},
"transitionStrategy": {
"regularTransitionStrategy": {
"type": "string",
"enum": ["disabled", "nvmesh", "rwxReadOnly"]
},
"helmTransitionStrategy": {
"type": "string",
"enum": ["disabled", "nvmesh"]
},
Expand All @@ -33,8 +37,8 @@
"additionalProperties": false,
"required": ["regularModelCache", "helmModelCache"],
"properties": {
"regularModelCache": {"$ref": "#/$defs/transitionStrategy"},
"helmModelCache": {"$ref": "#/$defs/transitionStrategy"}
"regularModelCache": {"$ref": "#/$defs/regularTransitionStrategy"},
"helmModelCache": {"$ref": "#/$defs/helmTransitionStrategy"}
}
},
"driver": {
Expand Down Expand Up @@ -79,17 +83,33 @@
}
}
}
},
{
"if": {
"properties": {
"transitions": {
"properties": {"regularModelCache": {"const": "rwxReadOnly"}},
"required": ["regularModelCache"]
}
},
"required": ["transitions"]
},
"then": {
"properties": {
"accessModes": {"contains": {"const": "ReadWriteMany"}}
}
}
}
]
},
"disabledDriver": {
"nonNVMeshDriver": {
"allOf": [
{"$ref": "#/$defs/driver"},
{
"properties": {
"transitions": {
"properties": {
"regularModelCache": {"const": "disabled"},
"regularModelCache": {"enum": ["disabled", "rwxReadOnly"]},
"helmModelCache": {"const": "disabled"}
}
}
Expand Down
3 changes: 2 additions & 1 deletion ...ane-services/nvca/deployments/nvca-operator/files/nvcf-storage-capabilities-v1alpha1.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,8 @@
# This public catalog is owned by NVCA and installed with the NVCA chart.
# accessModes lists only modes demonstrated for the exact tested provider.
# A disabled transition means that the end-to-end workflow is not qualified.
# The catalog is not yet wired into NVCA reconciliation or backend selection.
# NVCA reads this catalog when it creates a new model-cache storage selection.
# Only a registered, non-disabled transition can select durable caching.
apiVersion: storage.nvcf.nvidia.com/v1alpha1
kind: StorageCapabilityCatalog
drivers:
Expand Down
2 changes: 1 addition & 1 deletion ...pute-plane-services/nvca/deployments/nvca-operator/templates/pre-delete-cleanup-rbac.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -80,7 +80,7 @@ rules:
resources: ["icmsrequests/status"]
verbs: ["get", "update", "patch"]
- apiGroups: ["nvca.nvcf.nvidia.io"]
resources: ["storagerequests", "storagerequests/status"]
resources: ["modelcachebindings", "modelcachebindings/status", "storagerequests", "storagerequests/status"]
verbs: ["get", "list", "watch", "patch", "create", "update", "delete"]
- apiGroups: ["batch"]
resources: ["jobs", "cronjobs"]
Expand Down
2 changes: 1 addition & 1 deletion src/compute-plane-services/nvca/deployments/nvca-operator/templates/role.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,7 +63,7 @@ rules:
resources: ["icmsrequests/status"]
verbs: ["get", "update", "patch"]
- apiGroups: ["nvca.nvcf.nvidia.io"]
resources: ["storagerequests", "storagerequests/status"]
resources: ["modelcachebindings", "modelcachebindings/status", "storagerequests", "storagerequests/status"]
verbs: ["get", "list", "watch", "patch", "create", "update", "delete"]
- apiGroups: ["batch"]
resources: ["jobs", "cronjobs"]
Expand Down
Loading
Loading