ci: run the test matrix on the runner natively, not in a Debian container - #389
Merged
Conversation
…iner Every matrix job pulled a ruby:*-bullseye image and then apt-installed build tools, Chrome and Node (via NodeSource) from deb.debian.org. Those downloads were the source of the recurring 'Connection reset by peer' flakes that failed random single jobs across recent PRs. Switch to the standard runner-native pattern: - drop the container; run directly on ubuntu-latest - ruby/setup-ruby@v1 for the matrix Ruby (prebuilt binaries) - actions/setup-node@v6 for Node (drops the NodeSource apt step) - apt-install only the pg/rgeo-geos headers (libpq-dev, libgeos-dev) from Ubuntu's mirrors; gcc/make/patch and Chrome are preinstalled - point the test DB at 127.0.0.1: the postgis service is reachable by service name only from inside a job container; on the runner host it is published on localhost via the existing port mapping No change to what is tested (same Redmine/Ruby/PostGIS matrix, same suites); the host OS for the Ruby side becomes Ubuntu instead of Debian, which does not affect plugin-level tests.
There was a problem hiding this comment.
Pull request overview
Updates the PostGIS CI workflow to run the Ruby/Redmine/PostGIS test matrix directly on the GitHub-hosted Ubuntu runner (instead of inside a Debian-based Ruby container) to reduce network flakiness and simplify dependencies.
Changes:
- Removes the
container: ruby:*‑bullseyeconfiguration and installs Ruby viaruby/setup-ruby@v1. - Sets up Node via
actions/setup-node@v6and limits apt installs to native gem headers (libpq-dev,libgeos-dev). - Switches the DB host from the service name to
127.0.0.1to match runner-native service port publishing.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
- Build/test steps referenced the plugin dir as a literal redmine_gtt while
checkout uses ${{ env.PLUGIN_NAME }}; use the variable throughout so a
rename or fork doesn't break the build dir or test paths.
- release.yml pinned Node 22 while test-postgis.yml uses 24; align release on
24 so the frontend is built/tested/released on one Node major.
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
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.
Motivation
Every matrix job pulled a
ruby:*-bullseyeimage and then apt-installed build tools, Chrome, and Node (via NodeSource) fromdeb.debian.org. Those downloads were the source of the recurringError reading from server - read (104: Connection reset by peer)failures that flaked random single matrix jobs across the last several PRs (#383, #385, #386, #388), each needing a manual re-run.Change
Switch the
testjob to the standard runner-native GitHub Actions pattern, which removes almost all of that download surface:container: ruby:*-bullseye; run directly onubuntu-latest.ruby/setup-ruby@v1for the matrix Ruby (3.3 / 3.4) — prebuilt binaries, no apt.actions/setup-node@v6for Node — drops the NodeSource apt step (already used by thefrontendjob).libpq-devforpg,libgeos-devforrgeo-geos) from Ubuntu's Azure-hosted mirrors, which are far more reliable on GH runners thandeb.debian.org.gcc/make/patchand Chrome (for the system test) are preinstalled on the runner.127.0.0.1: thepostgisservice is reachable by service name only from inside a job container; on the runner host it's published onlocalhostvia the existingports: 5432:5432mapping. (This is the one functional gotcha of dropping the container.)Tradeoffs
No change to coverage: same Redmine/Ruby/PostGIS matrix, same unit/functional/integration/system suites, same uninstall test.