Repository navigation
Conversation
Adds a `--provision-cluster` path to `examples/xmc_lite_job/setup.sh` that creates a lightweight GKE cluster from scratch, so the example can run without a pre-existing cluster. - `xmc-lite-gke.yaml` is a Cluster Toolkit blueprint with a VPC, a GKE cluster, a 2x n2-standard-4 node pool, cert-manager, JobSet and Kueue. - `kueue-queues.yaml` adds a `default-flavor` ResourceFlavor, a `cluster-queue` ClusterQueue (6 CPU / 24Gi) and a `default-queue` LocalQueue, so `--local_queue=default-queue` works out of the box. - `setup.sh --provision-cluster` deploys the blueprint groups, creates the workload namespace if needed and applies the Kueue queues. - `README.md` lists the provisioning IAM roles (including `roles/iam.serviceAccountAdmin` and `roles/resourcemanager.projectIamAdmin`, needed to create the blueprint's service accounts and grant them project roles), how to build `gcluster`, and `gcluster destroy` teardown. Bug: 555533775 TAG=agy CONV=328157cc-0621-4429-b4c2-19760fa03506
There was a problem hiding this comment.
Code Review
This pull request adds automatic GKE cluster provisioning capabilities to the xmc_lite_job example using Cluster Toolkit (gcluster), including updating the setup script, adding a Cluster Toolkit blueprint, and defining Kueue resources. Feedback on these changes suggests adding a check in setup.sh to verify the existence of the gcluster binary before execution, and waiting for the Kueue controller manager deployment to be fully ready before applying Kueue queues to avoid webhook connection errors.
…-cluster # Conflicts: # examples/xmc_lite_job/README.md
- Wait for `deployment/kueue-controller-manager` to roll out before applying `kueue-queues.yaml`, so the queues are not rejected by Kueue webhooks that are not ready yet on a freshly provisioned cluster. - Fail early with build instructions when the `gcluster` binary is missing or not executable. Comment link: #1049 (comment) Comment link: #1049 (comment) Bug: 555533775 TAG=agy CONV=328157cc-0621-4429-b4c2-19760fa03506
…-cluster # Conflicts: # examples/xmc_lite_job/README.md
This branch has not been deployed
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
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
Adds
./setup.sh --provision-clustertoexamples/xmc_lite_job/: a Cluster Toolkit blueprint creates a lightweight GKE cluster with cert-manager, JobSet and Kueue, and ready-to-use Kueue queues are applied. PR 3 of 3 in the stack replacing #998.Motivation & Context
Extends the b/555533775 example for users without a cluster, providing the JobSet and Kueue setup needed by follow-up b/569821527. It also fixes gaps found when checking #998 against the story: missing provisioning IAM roles, no Kueue queues (so
--local_queuehad nothing to target), no instructions for gettinggcluster, and no cluster teardown.Key Changes
xmc-lite-gke.yaml: blueprint (VPC, GKE, 2xn2-standard-4node pool, cert-manager, Kueue 0.18.2, JobSet 0.12.0). Same as in feat(examples): add xmc lite kubernetes job and jobset example #998.kueue-queues.yaml(new):default-flavorResourceFlavor,cluster-queueClusterQueue (6 CPU / 24Gi),default-queueLocalQueue (kueue.x-k8s.io/v1beta2, matchingdeploy/helm/templates/kueue-dev).setup.sh:--provision-clusterdeploys the blueprint groups, creates${K8S_NAMESPACE}if needed and applies the queues.README.md: provisioning IAM roles, now includingroles/iam.serviceAccountAdminandroles/resourcemanager.projectIamAdmin(the blueprint createswl-sa/np-saand grants them project roles);gclusterbuild steps;gcluster destroyteardown.Assumptions
n2-standard-4pool, leaving headroom for system pods. If it's too low, workloads stay queued.Verification & Testing
bash -n setup.sh;./.githooks/pre-commit(pass).kueue-queues.yamlapply, namespace creation) has not been run against a fresh cluster yet.Related Issues / Bugs
Bug: 569821527
TAG=agy
CONV=328157cc-0621-4429-b4c2-19760fa03506