ci: Only keep relevant files after installing tests dependencies

The _build dir is 211MB (uncompressed) but the executables are moved to
the top gitaly folder after they're compiled, so there's no need to save
the _build dir in the artifacts of the setup-test-env job.

That should speed up the artifacts upload/download phases for
setup-test-env and all the jobs that depend on it.
Signed-off-by: Rémy Coutable <remy@rymai.me>

.gitlab/ci/dev-fixtures.gitlab-ci.yml

@@ -15,7 +15,6 @@
     # SEED_NESTED_GROUPS: "false" # requires network connection
 
 .run-dev-fixtures-script: &run-dev-fixtures-script
-  - run_timed_command "scripts/gitaly-test-build"
   - run_timed_command "scripts/gitaly-test-spawn"
   - run_timed_command "RAILS_ENV=test bundle exec rake db:seed_fu"
 

.gitlab/ci/frontend.gitlab-ci.yml

@@ -103,7 +103,6 @@ update-yarn-cache:
     WEBPACK_VENDOR_DLL: "true"
   script:
     - run_timed_command "gem install knapsack --no-document"
-    - run_timed_command "scripts/gitaly-test-build"
     - run_timed_command "scripts/gitaly-test-spawn"
     - source ./scripts/rspec_helpers.sh
     - rspec_paralellized_job "--tag frontend_fixture"

.gitlab/ci/rails.gitlab-ci.yml

@@ -10,7 +10,6 @@
   # Only install knapsack after bundle install! Otherwise oddly some native
   # gems could not be found under some circumstance. No idea why, hours wasted.
   - run_timed_command "gem install knapsack --no-document"
-  - run_timed_command "scripts/gitaly-test-build"
   - run_timed_command "scripts/gitaly-test-spawn"
   - source ./scripts/rspec_helpers.sh
 
@@ -150,20 +149,35 @@ setup-test-env:
   script:
     - run_timed_command "bundle exec ruby -I. -e 'require \"config/environment\"; TestEnv.init'"
     - run_timed_command "scripts/gitaly-test-build"  # Do not use 'bundle exec' here
-    - rm tmp/tests/gitaly/.ruby-bundle  # This file prevents gems from being installed even if vendor/gitaly-ruby is missing
   artifacts:
     expire_in: 7d
     paths:
       - config/secrets.yml
-      - tmp/tests/gitaly
-      - tmp/tests/gitlab-elasticsearch-indexer
-      - tmp/tests/gitlab-shell
-      - tmp/tests/gitlab-test-fork
-      - tmp/tests/gitlab-test-fork_bare
-      - tmp/tests/gitlab-test
-      - tmp/tests/gitlab-workhorse
-      - tmp/tests/repositories
-      - tmp/tests/second_storage
+      - tmp/tests/gitaly/config.toml
+      - tmp/tests/gitaly/gitaly
+      - tmp/tests/gitaly/gitaly2.config.toml
+      - tmp/tests/gitaly/gitaly-git2go
+      - tmp/tests/gitaly/gitaly-hooks
+      - tmp/tests/gitaly/gitaly-lfs-smudge
+      - tmp/tests/gitaly/gitaly-ssh
+      - tmp/tests/gitaly/internal/
+      - tmp/tests/gitaly/internal_sockets/
+      - tmp/tests/gitaly/Makefile
+      - tmp/tests/gitaly/praefect
+      - tmp/tests/gitaly/praefect.config.toml
+      - tmp/tests/gitaly/ruby/
+      - tmp/tests/gitlab-elasticsearch-indexer/bin/gitlab-elasticsearch-indexer
+      - tmp/tests/gitlab-shell/
+      - tmp/tests/gitlab-test-fork/
+      - tmp/tests/gitlab-test-fork_bare/
+      - tmp/tests/gitlab-test/
+      - tmp/tests/gitlab-workhorse/gitlab-zip-metadata
+      - tmp/tests/gitlab-workhorse/gitlab-zip-cat
+      - tmp/tests/gitlab-workhorse/gitlab-workhorse
+      - tmp/tests/gitlab-workhorse/gitlab-resize-image
+      - tmp/tests/gitlab-workhorse/config.toml
+      - tmp/tests/repositories/
+      - tmp/tests/second_storage/
     when: always
 
 update-rails-cache:

doc/development/pipelines.md

@@ -504,6 +504,10 @@ request, be sure to start the `dont-interrupt-me` job before pushing.
    - `update-yarn-cache`, defined in [`.gitlab/ci/frontend.gitlab-ci.yml`](https://gitlab.com/gitlab-org/gitlab/blob/master/.gitlab/ci/frontend.gitlab-ci.yml).
 1. These jobs run in merge requests whose title include `UPDATE CACHE`.
 
+### Artifacts strategy
+
+We limit the artifacts that are saved and retrieved by jobs to the minimum in order to reduce the upload/download time and costs, as well as the artifacts storage.
+
 ### Pre-clone step
 
 The `gitlab-org/gitlab` project on GitLab.com uses a [pre-clone step](https://gitlab.com/gitlab-org/gitlab/-/issues/39134)

scripts/gitaly-test-build

@@ -12,7 +12,7 @@ class GitalyTestBuild
   include GitalyTest
 
   def run
-    abort 'gitaly build failed' unless system(env, 'make', chdir: tmp_tests_gitaly_dir)
+    abort 'gitaly build failed' unless build_gitaly
 
     ensure_gitlab_shell_secret!
     check_gitaly_config!

scripts/gitaly-test-spawn

@@ -8,6 +8,7 @@ class GitalyTestSpawn
   include GitalyTest
 
   def run
+    install_gitaly_gems if ENV['CI']
     check_gitaly_config!
 
     # # Uncomment line below to see all gitaly logs merged into CI trace

scripts/gitaly_test.rb

@@ -41,7 +41,7 @@ module GitalyTest
       'HOME' => File.expand_path('tmp/tests'),
       'GEM_PATH' => Gem.path.join(':'),
       'BUNDLE_APP_CONFIG' => File.join(File.dirname(gemfile), '.bundle/config'),
-      'BUNDLE_FLAGS' => "--jobs=4 --retry=3 --quiet",
+      'BUNDLE_FLAGS' => "--jobs=4 --retry=3",
       'BUNDLE_INSTALL_FLAGS' => nil,
       'BUNDLE_GEMFILE' => gemfile,
       'RUBYOPT' => nil,
@@ -78,6 +78,14 @@ module GitalyTest
     end
   end
 
+  def install_gitaly_gems
+    system(env, "make #{tmp_tests_gitaly_dir}/.ruby-bundle", chdir: tmp_tests_gitaly_dir) # rubocop:disable GitlabSecurity/SystemCommandInjection
+  end
+
+  def build_gitaly
+    system(env, 'make', chdir: tmp_tests_gitaly_dir) # rubocop:disable GitlabSecurity/SystemCommandInjection
+  end
+
   def start_gitaly
     start(:gitaly)
   end

spec/support/capybara.rb

@@ -33,7 +33,7 @@ Capybara.register_server :puma_via_workhorse do |app, port, host, **options|
   socket_path = file.path
   file.close! # We just want the filename
 
-  TestEnv.with_workhorse(TestEnv.workhorse_dir, host, port, socket_path) do
+  TestEnv.with_workhorse(host, port, socket_path) do
     Capybara.servers[:puma].call(app, nil, socket_path, **options)
   end
 end

spec/support/helpers/test_env.rb

@@ -284,7 +284,7 @@ module TestEnv
     @workhorse_path ||= File.join('tmp', 'tests', 'gitlab-workhorse')
   end
 
-  def with_workhorse(workhorse_dir, host, port, upstream, &blk)
+  def with_workhorse(host, port, upstream, &blk)
     host = "[#{host}]" if host.include?(':')
     listen_addr = [host, port].join(':')