[TCL-6909] feat(slurm): Model-Aware TorchPass chart support (init scripts + orchestrator Service)#37
Open
sagrawal-byte wants to merge 3 commits into
Open
Conversation
Two additions, both gated/opt-in so default renders are unchanged: - login: support `login.initContainers` (the login Deployment template had no passthrough). Needed so the torchpass-payload initContainer can populate /opt/torchpass on login pods, matching the worker nodeset. - torchpass: new `torchpass.enabled` block rendering the cluster-scoped pieces the chart owns -- the torchpass-login-profile ConfigMap (sourced by login + worker shells) and the torchpass-orchestrator Service (:26001, selecting the slurmctld pod). The clockworks sidecar + payload initContainers + volumes are injected by the caller via the controller/nodeset/login podSpec values. Verified with `helm lint` + `helm template`. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
login (initScriptLogin) and compute (initScriptNodes) already mount a /root/init.sh init script via ConfigMap; the controller had no equivalent. Add initScriptController: a ConfigMap mounted at /root/init.sh on the slurmctld container, run at startup by the controller entrypoint (slurmcontainers controller.sh). Used to fetch role binaries (e.g. TorchPass) from the substrate metadata service. Gated/null by default -- no change to existing renders. Verified with helm lint + helm template. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ontroller' into sagrawal/torchpass-chart-pkg
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Combined TorchPass chart change (supersedes #35 + #36 — this is the exact source
tcloud-charts#354'sslurm-1.0.0.tgzwas packaged from):initScriptController— mounts/root/init.shon slurmctld via ConfigMap (mirrors the existinginitScriptLogin/initScriptNodes). Run by the controller entrypoint (slurmcontainers Add hostname services for NodeSets SlinkyProject/slurm-operator#50) to bootstrap TorchPass from the substrate metadata service.torchpass.enabledblock — renders thetorchpass-orchestratorService (:26001, selects the slurmctld pod) + login-profile ConfigMap.login.initContainerspassthrough (login Deployment previously had none).All gated/opt-in; default renders unchanged. Set by together-cloud-cluster-operator #658 when
Cluster.Spec.ModelAwareTorchPassEnabledis true.Test plan
helm lint→ 0 failures.helm templatewith all torchpass values: controller init CM +/root/init.shmount, orchestrator Service, login initContainer all render; default values render identical to base.test-slurm-torchpass(substrate-us-central-8a-2): metadata fetch → login stages+installs both roles → worker waits then SPANK-launches:26002→ controller deferred install + persisted strigger self-fires (monitor/plugin/orchestrator up on:26000/:26001).🤖 Generated with Claude Code