Draft
Conversation
Changelog: added - Documented use of Kyverno policies for Slurm-bridge
With exclusive allocations, all of the node's resources will report as being allocated to the Slurm job, hence the pod emit claims for all resources.
This following Kubernetes a little more closely. High level changes: * Move logic into patchPodExtendedResourceClaimStatus() * Move logic into bindClaim() * Rename createResourceClaim() to preBindExtendedResources() * Refactor generateRequestMappings() into createRequestsAndMappings() * Use established ResourceClaim.GenerateName schema * Use established ContainerExtendedResourceRequest name schema
Upgrade all modules. Need Kubernetes v1.35.x for DRA.
This module models Slurm nodes from Kubernetes nodes. Kube node ResourceSlices are parsed to assemble the representative Slurm node.
Similar to DRA for GPUs, we propagate the Slurm allocation for CPUs through extended resource claim. NOTES: * There are a number of magic values not exposed by the example CPU DRA driver which we need to use. * There are driver specific details to prepare the pod for CPU pinning. Changelog: Added - Added support for DRA CPU driver `dra-driver-cpu`.
Set controller.extraConfMap.ReconfigFlags to KeepPartInto during the deployment of the Slurm Helm Chart in order to keep the Slurm-bridge partition throughout reconfigures.
Add DRA CPU Support See merge request SchedMD/slinky-dev/slurm-bridge!110
Made-with: Cursor
feat: use KeepPartInfo in hack/kind.sh to prevent partition deletion See merge request SchedMD/slinky-dev/slurm-bridge!118
- Run only with a demo script with the examples - Change Makefile target name of "demo-dra" to "install-dra" - also removed lws from examples running, to not have the demo hang as they spin - update name of Makefile target for keys - update lws to 0.8.x - do not deploy dra items as part of demo-examples, that is a seperate target - Add demo-dra target that installs just the dra demo items
- also minor fix to yaml
- Add a special demo script that is better for a demo screen - Do not run watch script if on Mac - Remove unused code in Makefile - Fix memory block in kind script - Add demo targets for yamls only
Demo config See merge request SchedMD/slinky-dev/slurm-bridge!119
docs: document use of Kyverno policies with Slurm-bridge See merge request SchedMD/slinky-dev/slurm-bridge!115
Changelog: Fixed - Mutating webhook unsets nodeName, if pre-defined, to ensure that we schedule.
fix: Mutating webhook should unset spec.nodeName See merge request SchedMD/slinky-dev/slurm-bridge!120
cgetzen
force-pushed
the
fix/pod-deletion-race-condition
branch
from
66ec7fc to
11ec360
Compare
cgetzen
force-pushed
the
fix/pod-deletion-race-condition
branch
from
11ec360 to
47b0f88
Compare
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
4 participants
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.
Summary
Breaking Changes
Testing Notes
Additional Context