RFC: KEP-4381: DRA: optional allocation #4619
Open
+57
−0
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: pohly The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
k8s-ci-robot
added
the
kind/kep
k8s-ci-robot
added
cncf-cla: yes
pohly
commented
May 6, 2024
| selector: |- | ||
| attributes.quantity["memory"].isGreaterThan(quantity("32Gi")) | ||
| ``` | ||
|
|
There was a problem hiding this comment.
This by itself does not cover more complex requests like "Give me X of A, otherwise Y of B, or as a last resort Z of C". Having three requests with [0:X], [0:Y], [0:Z] is not sufficient because then too much (X, Y, and Z for each) or not enough could get allocated (0 of each).
For that we need some scoring mechanism that looks at all possible solutions and then evaluates which one is allowed and which one is best. Adding such scoring is the logical next step and would also be useful for other use cases.
asm582
reviewed
May 6, 2024
| # Each entry here is a request for one resource. | ||
| - namedresources: | ||
| range: | ||
| minimum: 0 |
There was a problem hiding this comment.
is it possible to have minimum == maximum?
There was a problem hiding this comment.
Yes. Both boundaries are inclusive (min <= x <= max).
There was a problem hiding this comment.
As an example or in the NamedResourcesRange definition below?
There was a problem hiding this comment.
I vote for adding it in NamedResourcesRange definition
pohly
changed the title
bart0sh
reviewed
May 7, 2024
| @@ -819,6 +820,42 @@ namedResources: | |||
| - gpu-1 | |||
| ``` | |||
|
|
|||
| ### Optional allocation | |||
|
|
|||
| The default used by the examples above is to allocated exactly one named | |||
There was a problem hiding this comment.
typo: allocated -> allocate
bart0sh
reviewed
May 7, 2024
| will be listed in the allocation result, same as when no range is | ||
| specified. That list may be empty if nothing was allocated. | ||
|
|
||
| This example asks for all GPUs with more than 32Gi that are available: |
There was a problem hiding this comment.
Suggested change
| This example asks for all GPUs with more than 32Gi that are available: | |
| This example asks for all GPUs with more than 32Gi of memory that are available: |
pohly
force-pushed
the
dra-optional-allocation
branch
from
a701b0f
to
a65a12c
Compare
Running a pod without some resource may be useful if there is a fallback in the application code. Allocating more than one may also be useful, in particular when the upper bound is unlimited. Then it becomes possible to ask for all instances available on a node in a daemon set.
pohly
force-pushed
the
dra-optional-allocation
branch
from
a65a12c
to
f852965
Compare
pohly
added a commit
to pohly/wg-device-management
that referenced
this pull request
May 13, 2024
This proposal takes the existing KEP as base and includes: - vendor-independent classes and attributes (kubernetes/enhancements#4614) - optional allocation (kubernetes/enhancements#4619) - inline parameters (kubernetes/enhancements#4613) - management access (kubernetes/enhancements#4611) - renaming "named resources" to "devices" wherever it makes sense and is user-facing (Slack discussion) - MatchAttributes (from k8srm-prototype) - OneOf (from k8srm-prototype) `pkg/api` currently builds, but the rest doesn't. None of the YAML examples have been updated yet.
pohly
added a commit
to pohly/wg-device-management
that referenced
this pull request
May 13, 2024
This proposal takes the existing KEP as base and includes: - vendor-independent classes and attributes (kubernetes/enhancements#4614) - optional allocation (kubernetes/enhancements#4619) - inline parameters (kubernetes/enhancements#4613) - management access (kubernetes/enhancements#4611) - renaming "named resources" to "devices" wherever it makes sense and is user-facing (Slack discussion) - MatchAttributes (from k8srm-prototype) - OneOf (from k8srm-prototype) `pkg/api` currently builds, but the rest doesn't. None of the YAML examples have been updated yet.
pohly
added a commit
to pohly/wg-device-management
that referenced
this pull request
May 13, 2024
This proposal takes the existing KEP as base and includes: - vendor-independent classes and attributes (kubernetes/enhancements#4614) - optional allocation (kubernetes/enhancements#4619) - inline parameters (kubernetes/enhancements#4613) - management access (kubernetes/enhancements#4611) - renaming "named resources" to "devices" wherever it makes sense and is user-facing (Slack discussion) - MatchAttributes (from k8srm-prototype) - OneOf (from k8srm-prototype) `pkg/api` currently builds, but the rest doesn't. None of the YAML examples have been updated yet.
pohly
added a commit
to pohly/wg-device-management
that referenced
this pull request
May 13, 2024
This proposal takes the existing KEP as base and includes: - vendor-independent classes and attributes (kubernetes/enhancements#4614) - optional allocation (kubernetes/enhancements#4619) - inline parameters (kubernetes/enhancements#4613) - management access (kubernetes/enhancements#4611) - renaming "named resources" to "devices" wherever it makes sense and is user-facing (Slack discussion) - MatchAttributes (from k8srm-prototype) - OneOf (from k8srm-prototype) `pkg/api` currently builds, but the rest doesn't. None of the YAML examples have been updated yet.
pohly
added a commit
to pohly/wg-device-management
that referenced
this pull request
May 16, 2024
This proposal takes the existing KEP as base and includes: - vendor-independent classes and attributes (kubernetes/enhancements#4614) - optional allocation (kubernetes/enhancements#4619) - inline parameters (kubernetes/enhancements#4613) - management access (kubernetes/enhancements#4611) - renaming "named resources" to "devices" wherever it makes sense and is user-facing (Slack discussion) - MatchAttributes (from k8srm-prototype) - OneOf (from k8srm-prototype) `pkg/api` currently builds, but the rest doesn't. None of the YAML examples have been updated yet.
pohly
added a commit
to pohly/wg-device-management
that referenced
this pull request
May 16, 2024
This proposal takes the existing KEP as base and includes: - vendor-independent classes and attributes (kubernetes/enhancements#4614) - optional allocation (kubernetes/enhancements#4619) - inline parameters (kubernetes/enhancements#4613) - management access (kubernetes/enhancements#4611) - renaming "named resources" to "devices" wherever it makes sense and is user-facing (Slack discussion) - MatchAttributes (from k8srm-prototype) - OneOf (from k8srm-prototype) `pkg/api` currently builds, but the rest doesn't. None of the YAML examples have been updated yet.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
One-line PR description: optional allocation
Issue link: DRA: structured parameters #4381
Other comments:
Running a pod without some resource may be useful if there is a fallback in the application code. Allocating more than one may also be useful, in particular when the upper bound is unlimited. Then it becomes possible to ask for all instances available on a node in a daemon set.