mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 12:21:27 +01:00
e5c0e6b5e3
I think we usually we leave the ` / test` in for clarity Pull Request resolved: https://github.com/pytorch/pytorch/pull/164287 Approved by: https://github.com/atalman, https://github.com/malfet |
||
|---|---|---|
| .. | ||
| upload_utilization_stats | ||
| __init__.py | ||
| check_disabled_tests.py | ||
| export_test_times.py | ||
| import_test_stats.py | ||
| monitor.py | ||
| README.md | ||
| sccache_stats_to_benchmark_format.py | ||
| test_dashboard.py | ||
| upload_artifacts.py | ||
| upload_dynamo_perf_stats.py | ||
| upload_external_contrib_stats.py | ||
| upload_metrics.py | ||
| upload_sccache_stats.py | ||
| upload_stats_lib.py | ||
| upload_test_stats_intermediate.py | ||
| upload_test_stats_running_jobs.py | ||
| upload_test_stats.py | ||
| utilization_stats_lib.py | ||
PyTorch CI Stats
We track various stats about each CI job.
- Jobs upload their artifacts to an intermediate data store (either GitHub
Actions artifacts or S3, depending on what permissions the job has). Example:
a9f6a35a33/.github/workflows/_linux-build.yml (L144-L151) - When a workflow completes, a
workflow_runevent triggersupload-test-stats.yml. upload-test-statsdownloads the raw stats from the intermediate data store and uploads them as JSON to s3, which then uploads to our database backend
graph LR
J1[Job with AWS creds<br>e.g. linux, win] --raw stats--> S3[(AWS S3)]
J2[Job w/o AWS creds<br>e.g. mac] --raw stats--> GHA[(GH artifacts)]
S3 --> uts[upload-test-stats.yml]
GHA --> uts
uts --json--> s3[(s3)]
s3 --> DB[(database)]
Why this weird indirection? Because writing to the database requires special permissions which, for security reasons, we do not want to give to pull request CI. Instead, we implemented GitHub's recommended pattern for cases like this.
For more details about what stats we export, check out
upload-test-stats.yml