Add blueprint for ml diagnostics#5350
Add blueprint for ml diagnostics#5350AdarshK15 wants to merge 5 commits intoGoogleCloudPlatform:developfrom
Conversation
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly enhances the Cluster Toolkit by introducing a new blueprint that streamlines the deployment of GKE clusters tailored for high-performance AI/ML tasks using TPU v6e accelerators. It simplifies the complex setup of networking, IAM, and Kubernetes scheduling components, while also pre-integrating Google Cloud ML Diagnostics to ensure workloads are immediately observable and diagnosable. This change aims to reduce operational overhead and accelerate the development and deployment of machine learning applications on Google Cloud. Highlights
Changelog
Activity
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a new blueprint for ML Diagnostics on GKE with TPUs. It adds a new example, a new mldiagnostics Terraform module, and supporting changes to the kubectl-apply and gke-cluster modules. The changes are well-structured and the new wait output in kubectl-apply is a good pattern for explicit dependencies. However, I've identified a few issues, primarily in the new mldiagnostics module related to incorrect namespace handling and dependency definitions which could cause deployment failures. I've also found some inconsistencies and a typo in the new example's documentation and sample job. My review includes detailed comments and code suggestions to address these points.
examples/gke-ml-diagnostics/gke-ml-diagnostics-tpu-v6e-deployment.yaml
Outdated
Show resolved
Hide resolved
|
Please add a PR description |
examples/gke-ml-diagnostics/gke-ml-diagnostics-tpu-v6e-deployment.yaml
Outdated
Show resolved
Hide resolved
| release_channel: REGULAR | ||
| maintenance_exclusions: | ||
| - name: no-minor-or-node-upgrades-indefinite | ||
| start_time: "2025-12-01T00:00:00Z" |
There was a problem hiding this comment.
why is this hardcoded and outed datetime?
| # See the License for the specific language governing permissions and | ||
| # limitations under the License. | ||
|
|
||
| blueprint_name: gke-ml-diagnostics-tpu |
There was a problem hiding this comment.
please add daily test for same to make sure the blueprint is working all time
| source: modules/management/mldiagnostics | ||
| use: [gke-tpu-v6e-cluster, workload-manager-install] | ||
| settings: | ||
| workload_manager_wait: $(workload-manager-install.wait) |
There was a problem hiding this comment.
why not add it to the use clause?
| use: [gke-tpu-v6e-cluster, workload-manager-install] | ||
| settings: | ||
| workload_manager_wait: $(workload-manager-install.wait) | ||
| namespace: $(vars.namespace) |
There was a problem hiding this comment.
is the namespace going to clash with xpk?
| ip_cidr_range: 10.0.32.0/20 | ||
| firewall_rules: | ||
| - name: $(vars.deployment_name)-internal-0 | ||
| ranges: [192.168.0.0/16] |
There was a problem hiding this comment.
Shouldn't this range match with the subnet ip?
| subnet_ip: 192.168.64.0/18 | ||
| firewall_rules: | ||
| - name: $(vars.deployment_name)-internal-1 | ||
| ranges: [192.168.0.0/16] |
| - artifactregistry.reader | ||
| - hypercomputecluster.editor | ||
| - storage.admin | ||
| # - storage.objectUser |
There was a problem hiding this comment.
if not used, should we remove this?
3 participants
Submission Checklist
NOTE: Community submissions can take up to 2 weeks to be reviewed.
Please take the following actions before submitting this pull request.