Add unallocated_resources option to Compute.get_offers()#4067
Open
un-def wants to merge 1 commit into
Open
Conversation
Adds a new `unallocated_resources` argument alongside `full_offers`. When set, backends that pack multiple jobs onto a node (Kubernetes; not yet Slurm) subtract already-allocated resources so offers reflect only unallocated (available) capacity. Exposed as the `--unallocated` CLI flag (`dstack offer` and `dstack apply`) and the `unallocated_resources` API parameter of `get_plan`/`list_gpus`. It is a discovery-only knob: the scheduling path always passes `False`. Kubernetes previously subtracted allocated resources unconditionally. That is now off: the cached allocation view can lag reality (e.g. a node whose previous job already terminated still looks allocated), which blocked provisioning of nodes that were actually free. Provisioning is now optimistic and lets the scheduler decide; the short `JOB_POD_SCHEDULING_TIMEOUT` makes a retry cheap.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a new
unallocated_resourcesargument alongsidefull_offers. When set, backends that pack multiple jobs onto a node (Kubernetes; not yet Slurm) subtract already-allocated resources so offers reflect only unallocated (available) capacity. Exposed as the--unallocatedCLI flag (dstack offeranddstack apply) and theunallocated_resourcesAPI parameter ofget_plan/list_gpus. It is a discovery-only knob: the scheduling path always passesFalse.Kubernetes previously subtracted allocated resources unconditionally. That is now off: the cached allocation view can lag reality (e.g. a node whose previous job already terminated still looks allocated), which blocked provisioning of nodes that were actually free. Provisioning is now optimistic and lets the scheduler decide; the short
JOB_POD_SCHEDULING_TIMEOUTmakes a retry cheap.