Minimizing flakiness of dispatcher_test.rb

[p-schlickmann]

Issue: #602
Flake reference: https://github.com/rails/solid_queue/blob/main/test/unit/dispatcher_test.rb
Failed run: https://github.com/rails/solid_queue/actions/runs/18917303921/job/54003968729

Summary

1. polling queries are logged — Rewind StringIO and increase sleep
2. polling queries can be silenced — Rewind StringIO and increase sleep
3. silencing polling queries when there's no Active Record logger — Increase sleep
4. run more than one instance of the dispatcher — Increase AddToBufferJob wait and increase sleep
5. sleeps 0.seconds between polls if there are ready to dispatch jobs — Increase AddToBufferJob wait and increase sleep

While the CI logs helped diagnose flakes in tests 1 and 2, the following command was useful to reproduce others locally:

export TARGET_DB=sqlite
export BUNDLE_GEMFILE=./gemfiles/rails_7_1.gemfile # or whatever
export RAILS_ENV=test
# Run 4000 tests in parallel using 4 CPUs (you must install `parallel` CLI)
seq 1 4000 | parallel -j4 --halt now,fail=1 'TEST_ENV_NUMBER={%} bin/rails test test/unit/dispatcher_test.rb'

You might also need to add TEST_ENV_NUMBER to database.yml:

test:
  primary:
    <<: *default
    database: <%= database_name_from("test" + ENV["TEST_ENV_NUMBER"]) %>
  shard_one:
    <<: *default
    database: <%= database_name_from("test_shard_one" + ENV["TEST_ENV_NUMBER"]) %>
    migrations_paths: db/migrate_shards
  shard_two:
    <<: *default
    database: <%= database_name_from("test_shard_two" + ENV["TEST_ENV_NUMBER"]) %>
    migrations_paths: db/migrate_shards
  queue:
    <<: *default
    database: <%= database_name_from("test_queue" + ENV["TEST_ENV_NUMBER"]) %>
    migrations_paths: db/queue_migrate

And also modify the log path for environments/test.rb:

Rails.application.configure do
    # ...
    config.paths["log"] = "log/test#{ENV['TEST_ENV_NUMBER']}.log"
end

[rosa]

Awesome @p-schlickmann! 🙇🏻‍♀️