Skip to content

Remove support for nvidia driver capabilities#798

Closed
elezar wants to merge 1 commit intocompose-spec:mainfrom
elezar:remove-nvidia-capabilities
Closed

Remove support for nvidia driver capabilities#798
elezar wants to merge 1 commit intocompose-spec:mainfrom
elezar:remove-nvidia-capabilities

Conversation

@elezar
Copy link

@elezar elezar commented Jun 18, 2025

This chanage updates the device spec to allow for a driver OR capabilities to be specified. It also removes the capabilities from the gpu spec and indicates that the gpu capability is implied.

@elezar elezar requested a review from ndeloof as a code owner June 18, 2025 16:16
@elezar elezar mentioned this pull request Jun 18, 2025
Draft
@ndeloof
Copy link
Collaborator

ndeloof commented Jun 18, 2025

It also removes the capabilities from the gpu spec and indicates that the gpu capability is implied.

true, still other capabilities could be required to select the adequate device.
Also, being version-less, compose-spec doesn't allow attributes to be removed from the spec.

@elezar
Remove support for nvidia driver capabilities
9a8a484 
This chanage updates the device spec to allow for a driver
OR capabilities to be specified. It also removes the NVIDIA capabilitied
from the documentation of the gpus.capabilities field.

Signed-off-by: Evan Lezar <elezar@nvidia.com>
@elezar elezar force-pushed the remove-nvidia-capabilities branch from d498a2d to 9a8a484 Compare June 18, 2025 20:22
@elezar
Copy link
Author

elezar commented Jun 18, 2025

It also removes the capabilities from the gpu spec and indicates that the gpu capability is implied.

true, still other capabilities could be required to select the adequate device. Also, being version-less, compose-spec doesn't allow attributes to be removed from the spec.

Thanks for pointing this out. I have updated the PR to not remove the field, but to update the documentation for the field to not refer to the nvidia-specific capabilities.

@ndeloof ndeloof enabled auto-merge (rebase) June 19, 2025 06:51
@vvoland
Copy link
Contributor

vvoland commented Jun 26, 2025

@elezar looks like the failure is related:

 --- FAIL: TestServiceDeviceRequestCountIntegerType (0.01s)
    loader_test.go:2164: assertion failed: error is not nil: validating filename0.yml: services.hello-world.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched
--- FAIL: TestServiceDeviceRequestCountStringType (0.01s)
    loader_test.go:2181: assertion failed: error is not nil: validating filename0.yml: services.hello-world.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched
--- FAIL: TestServiceDeviceRequestCountIntegerAsStringType (0.01s)
    loader_test.go:2199: assertion failed: error is not nil: validating filename0.yml: services.hello-world.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched
--- FAIL: TestServiceDeviceRequestWithoutCountAndDeviceIdsType (0.01s)
    loader_test.go:2215: assertion failed: error is not nil: validating filename0.yml: services.hello-world.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched
--- FAIL: TestServiceDeviceRequestCountInvalidStringType (0.01s)
    loader_test.go:2233: assertion failed: expected error to contain "invalid value \"some_string\", the only value allowed is 'all' or a number", got "validating filename0.yml: services.hello-world.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched"
--- FAIL: TestServiceDeviceRequestCountAndDeviceIdsExclusive (0.01s)
    loader_test.go:2251: assertion failed: expected error to contain "\"count\" and \"device_ids\" attributes are exclusive", got "validating filename0.yml: services.hello-world.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched"
--- FAIL: TestServiceDeviceRequestCapabilitiesMandatory (0.01s)
    loader_test.go:2267: assertion failed: expected an error, got nil
time="2025-06-19T06:51:56Z" level=warning msg="\"yes\" for boolean is not supported by YAML 1.2, please use `true`"
--- FAIL: TestLoadDeviceReservation (0.00s)
    loader_test.go:3615: assertion failed: error is not nil: validating filename0.yml: services.test.deploy.resources.reservations.devices.0 oneOf failed, subschemas 0, 1 matched
time="2025-06-19T06:51:57Z" level=warning msg="The \"UNSET_VAR\" variable is not set. Defaulting to a blank string."
FAIL

@github-actions github-actions bot added the Stale label Aug 27, 2025
@github-actions github-actions bot closed this Sep 7, 2025
auto-merge was automatically disabled September 7, 2025 00:19

Pull request was closed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ndeloof ndeloof ndeloof approved these changes

Assignees

No one assigned

Labels

Stale

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@elezar @ndeloof @vvoland