ci: add GPU runner overlay2 smoke test workflow and runner config sample - #7814
Open
MrLi000001 wants to merge 3 commits into
Open
ci: add GPU runner overlay2 smoke test workflow and runner config sample#7814MrLi000001 wants to merge 3 commits into
MrLi000001 wants to merge 3 commits into
Conversation
- gpu-runner-overlay2-smoke.yml: workflow_dispatch probe validating Docker storage driver (overlay2), cold/warm image pull timings, and nvidia-smi GPU visibility inside ghcr.io/deepmodeling/abacus-cuda image - cpu-runner-overlay2-sample.yaml: K8s RunnerDeployment sample (ACK/ECI) switching docker daemon storage driver from vfs to overlay2 Note: smoke workflow expects overlay2; on the current vfs runner it will fail at the storage-driver check by design.
The previous commit placed both the smoke workflow and the K8s RunnerDeployment sample under .github/workflows/. GitHub Actions parses every YAML in that directory as a workflow, which caused the K8s sample to be interpreted as a broken workflow (run deepmodeling#1 failed at parse step, shadowing the real smoke workflow in the Actions UI). Move the K8s sample to docs/community/cicd/overlay2-runner-sample.yaml where it belongs as CI documentation, not as an Actions workflow.
mohanchen
added
the
Compile & CICD & Docs & Dependencies
Add pull_request trigger scoped to the smoke workflow file itself, so updates to the workflow (e.g. validating changes to the runner probe) are automatically validated instead of requiring a manual dispatch. Keeps workflow_dispatch for on-demand runs.
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Adds a workflow_dispatch smoke test that validates the GPU self-hosted runner setup for Docker overlay2 storage driver, plus a K8s RunnerDeployment sample showing how to switch the runner daemon from vfs to overlay2.
Files
.github/workflows/gpu-runner-overlay2-smoke.ymloverlay2, measures cold/warm pull time ofghcr.io/deepmodeling/abacus-cuda:latest, runsnvidia-smiinside the image, and reports disk usage + timings to the job summarydocs/community/cicd/overlay2-runner-sample.yamldaemon.jsonwithstorage-driver: overlay2(ConfigMap + emptyDir docker storage)How to use
Docker driver: overlay2, cold/warm pull timings, GPU test passed.Context
This is a diagnostic companion for the CI optimization series (#7753 cuda-arch-parallel, etc.): a slow vfs image layer unpack is suspected to contribute to CUDA build/test runtime variance on the self-hosted GPU runners. This smoke test quantifies the pull cost before/after the overlay2 migration.