Repository navigation
Conversation
There was a problem hiding this comment.
Claude Code Review
This repository is configured for manual code reviews. Comment @claude review for a one-time review, or @claude review always to subscribe this PR to a review on every future push.
Tip: disable this comment in your organization's Code Review settings.
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
✅ Gate recommendation: approve.
Using the wrapper override alongside Maven’s central mirror covers both bootstrap downloads and plugin/dependency resolution. The configuration stays within the three CI workflows and preserves their existing commands, matrices, caching, and permissions.
Motivation
CI can fail before compilation or tests when Maven Central returns errors while downloading Maven, plugins, or dependencies. Earlier runs hit HTTP 429 for
maven-source-plugin:2.2.1; a controlled cold-cache comparison also reproduced HTTP 404 and 429 while downloading Maven itself.Configure the Google-hosted Maven Central mirror through
setup-java's existingmvn-mirrorsinput. SetMVNW_REPOURLfor the wrapper download, which happens before Maven readssettings.xml, following Trino's CI setup.The final change adds 10 lines across the Spark, AWS, and Search workflows. Cache behavior, dependency versions, build commands, timeouts, and job parallelism are unchanged.
Validation
Controlled comparison: three repetitions of the 16 Spark workflow tasks per source, using the same application code (
c0fbc96), runner type, JDK, commands, and timeouts. Each task used new Maven dependency and wrapper directories, with Maven cache restore/save disabled. All 96 cold-cache and source checks passed. Source pairs ran in one temporary workflow with at most 32 concurrent jobs; failed jobs were not rerun.All seven failures occurred while downloading Maven 3.9.9: six HTTP 404 responses and one HTTP 429 response. Every other task passed; there were no test failures or timeouts. Thus all three original-source groups failed, while all three mirror groups passed.
This directly supports the wrapper source override. The comparison does not isolate the dependency/plugin mirror's independent benefit or establish a long-term failure rate. The experiment workflow is on a separate branch and is not part of this PR.
actionlintandgit diff --checkpassed.LanceSparkWriteOptionsTestcases passed.