ci: parallelise tests-providers into a per-provider matrix#725
Merged
Conversation
The four diagnostic providers ran serially in one runner (~10m44s, dominated by esmvaltool at ~5m). Split them into a matrix so they run in parallel on separate runners; the providers critical path drops to the slowest single provider (~6m). - Each job fetches only sample data up front; the ilamb target fetches its own ilamb-test registry, so the other providers skip that download. - Per-job `coverage report` is non-gating (it is scoped to a single provider); the aggregate fail_under gate is enforced by codecov/project, since all four jobs upload to the shared `providers` flag. - Scratch artifacts are named per provider to avoid upload-artifact collisions.
Codecov Report✅ All modified and coverable lines are covered by tests.
Flags with carried forward coverage won't be shown. Click here to find out more. 🚀 New features to boost your workflow:
|
Split the long tests-core job (core + ref + celery + integration run serially, ~11m) into two parallel jobs: tests-core (core, celery and the climate-ref unit tests) and a new tests-integration job (the climate-ref integration suite plus the top-level cross-package tests). Both upload to the `core` codecov flag with `--fail-under=0`; the aggregate gate stays on codecov/project. Add `concurrency` groups to ci, packaging and provider-compat so superseded pull-request runs are cancelled (main/tags are never cancelled), and trim the provider-compatibility matrix to a single Python version on pull requests. Together these keep the per-PR job fan-out under the runner concurrency cap while roughly halving the critical path. The Makefile gains test-ref-unit / test-ref-integration targets (test-ref runs both); the unit target keeps --doctest-modules so src module-level coverage is still tracked. Also harden a providers-list test whose assertion depended on a long tmp_path not being truncated in the rendered rich table, which fails under pytest-xdist.
`coverage xml` re-applies the pyproject `fail_under=70` after writing the report, so the tests-integration job failed (partial coverage ~49%) even though `coverage report` was already run with `--fail-under=0`. Pass `--fail-under=0` to `coverage xml` too; the aggregate gate is enforced by codecov/project. Same latent issue applied to the tests-core and tests-providers jobs, fixed consistently.
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.
Description
Follow-up to #723. The
tests-providersCI job ran all four diagnostic providers serially in a single runner, which made it the critical path for the CI.This splits it into a
matrixover[example, esmvaltool, ilamb, pmp]so the providers run in parallel on separate runners.The integration tests are also split out from the unit tests
Checklist
Please confirm that this pull request has done the following:
changelog/