Scheduled maintenance pipeline for Ruby 3.1 broken
I set up nightly Ruby 3.1 pipeline schedule to prepare !140686 (merged) at https://gitlab.com/gitlab-org/gitlab-foss/-/pipeline_schedules
However a lot of RSpec jobs have an error like this: https://gitlab.com/gitlab-org/gitlab-foss/-/jobs/5831996304
An error occurred while loading spec_helper.
Failure/Error: config.register Tooling::Crystalball::CoverageLinesStrategy.new(execution_detector)
RuntimeError:
coverage measurement is already setup
# ./tooling/lib/tooling/crystalball/coverage_lines_strategy.rb:19:in `start'
# ./tooling/lib/tooling/crystalball/coverage_lines_strategy.rb:19:in `after_register'
# ./vendor/ruby/3.1.0/gems/crystalball-0.7.0/lib/crystalball/map_generator/configuration.rb:54:in `register'
# ./spec/crystalball_env.rb:22:in `block in start!'
# ./vendor/ruby/3.1.0/gems/crystalball-0.7.0/lib/crystalball/map_generator.rb:31:in `initialize'
# ./vendor/ruby/3.1.0/gems/crystalball-0.7.0/lib/crystalball/map_generator.rb:16:in `new'
# ./vendor/ruby/3.1.0/gems/crystalball-0.7.0/lib/crystalball/map_generator.rb:16:in `start!'
# ./spec/crystalball_env.rb:20:in `start!'
# ./spec/spec_helper.rb:22:in `<top (required)>'
It's fine for Ruby 3.0 though: https://gitlab.com/gitlab-org/gitlab-foss/-/jobs/5829483179
Clearly this isn't a problem for EE but only FOSS. Why? How can we fix it?
It's also broken for EE: https://gitlab.com/gitlab-org/gitlab/-/pipelines/1125867755
Edited by Lin Jen-Shin