OSDOCS Node DRA: Consumable Capacity GA enablement - #119614
mburke5678 wants to merge 7 commits into
Conversation
openshift-ci
Bot
added
the
size/M
|
🤖 Mon Sep 14 15:31:36 - Prow CI generated the docs preview: |
openshift-ci
Bot
added
size/L
| + | ||
| When a resource claim is created and used in a pod, {product-title} uses the resource slices to find nodes that have the requested resources. After finding an eligible resource slice for the resource claim, the {product-title} scheduler updates the resource claim with the allocation details, allocates resources to the resource claim, and schedules the pod onto a node that can access the resources. | ||
| + | ||
| You can define these slices by partitioning a device explicitly or allow the {product-title} scheduler to automatically define the slices. |
| + | ||
| Logical devices can specify the `consumesCounters` list in a resource slice. Each entry contains a reference to a counter set and a set of named counters with the amounts they can consume. | ||
| + | ||
| In the following example, the `resource-driver.example.com` DRA driver is sharing 40Gi of memory through the `gpu-0-counter-set` counter set in the `counter-slice` object. The `device-slice` object makes this memory available as a full device named _gpu-0_ or either of two partitioned devices named _gpu-0-partition-0_ and _gpu-0-partition-1_. Each device has a capacity and declares how much it consumes from the counter when allocated, either 40Gi for the full device or 20Gi of memory for each partition. You can allocate the memory by using either the full device or the partitions, but not both, enforcing mutual exclusivity between a full device and its partitions. When _gpu-0_ is allocated, the GPU consumes all 40Gi from the counter set, leaving 0Gi for the partitions. |
There was a problem hiding this comment.
This text is existing. I added it to a definition list.
| `spec.devices.consumesCounters`:: Specifies how much of a shared counter set the device consumes if allocated. | ||
| + | ||
| -- | ||
| + |
There was a problem hiding this comment.
The following text is all new.
| + | ||
| -- | ||
| * *Resource template claim for partitioned devices* | ||
| + |
There was a problem hiding this comment.
The following is existing text added to a definition list.
|
|
||
| `spec.devices.requests.exactly.selectors` or `spec.devices.requests.firstAvailable.selectors`:: Specifies CEL expressions to request specific devices from the specified device class. | ||
| -- | ||
| + |
There was a problem hiding this comment.
The following text is new.
|
@bitoku Can you PTAL? Does this cover what the user needs in order to use the consumable capacity DRA feature? |
| + | ||
| When a resource claim is created and used in a pod, {product-title} uses the resource slices to find nodes that have the requested resources. After finding an eligible resource slice for the resource claim, the {product-title} scheduler updates the resource claim with the allocation details, allocates resources to the resource claim, and schedules the pod onto a node that can access the resources. | ||
| + | ||
| You can define these slices by partitioning a device explicitly or allow the {product-title} scheduler to automatically define the slices. |
There was a problem hiding this comment.
Is this doc for users or dra driver developers?
DRA driver publishes resource slices, so users normally can't define slices.
Also the scheduler doesn't define slices. It just manages how much the capacity is consumed.
The consumable capacity feature allows the same devices to be consumed by multiple independent ResourceClaims, with the Kubernetes scheduler managing how much of the device's capacity is used up by each claim.
There was a problem hiding this comment.
The docs are for users not DRA driver developers. I wanted to show what folks can do with DRA and what the objects look like.
|
@mburke5678: all tests passed! Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here. |
|
@BhargaviGudi PTAL |
4 participants
https://redhat.atlassian.net/browse/OSDOCS-22103
Link to docs preview:
QE review: