On Mon, 21 Oct 2024 14:41:48 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
> When GH output is too large, we do not actually truncate it after > [JDK-8341424](https://bugs.openjdk.org/browse/JDK-8341424). The error log > would be: > > > Run bash ./.github/scripts/gen-test-results.sh "$GITHUB_STEP_SUMMARY" > ./.github/scripts/gen-test-results.sh: line 28: report-utils.sh: No such file > or directory > ./.github/scripts/gen-test-results.sh: line 97: truncate_summary: command not > found > > Error: Process completed with exit code 127. > Error: $GITHUB_STEP_SUMMARY upload aborted, supports content up to a size of > 1024k, got 1579k. For more information see: > https://docs.github.com/actions/using-workflows/workflow-commands-for-github-actions#adding-a-markdown-summary > > > This is because we import report-utils.sh incorrectly: scripts try to find it > in current directory, which is not the same directory where the importing > scripts are located. And we never see it in normal cases, because scripts are > just eating the errors without any other observable effects. > > Additional testing: > - [x] GHA checks work with deliberately failing tests Marked as reviewed by erikj (Reviewer). ------------- PR Review: https://git.openjdk.org/jdk/pull/21617#pullrequestreview-2382634121
