gitlab-org--gitlab-foss/spec/services/groups/destroy_service_spec.rb
Valery Sizov 15d011d64d Geo: sync disabled wikis. Stage 2
We started syncing all the wiki regardless of the fact it's disabled or
not. We couldn't do that in one stage because of needing of smoth update
and deprecating things. This is the second stage that finally removes
unused columns in the geo_node_status table.
2018-09-25 17:49:39 +03:00

170 lines
5.7 KiB
Ruby

require 'spec_helper'
describe Groups::DestroyService do
include DatabaseConnectionHelpers
let!(:user) { create(:user) }
let!(:group) { create(:group) }
let!(:nested_group) { create(:group, parent: group) }
let!(:project) { create(:project, :legacy_storage, namespace: group) }
let!(:notification_setting) { create(:notification_setting, source: group)}
let(:gitlab_shell) { Gitlab::Shell.new }
let(:remove_path) { group.path + "+#{group.id}+deleted" }
before do
group.add_user(user, Gitlab::Access::OWNER)
end
def destroy_group(group, user, async)
if async
Groups::DestroyService.new(group, user).async_execute
else
Groups::DestroyService.new(group, user).execute
end
end
shared_examples 'group destruction' do |async|
context 'database records' do
before do
destroy_group(group, user, async)
end
it { expect(Group.unscoped.all).not_to include(group) }
it { expect(Group.unscoped.all).not_to include(nested_group) }
it { expect(Project.unscoped.all).not_to include(project) }
it { expect(NotificationSetting.unscoped.all).not_to include(notification_setting) }
end
context 'mattermost team' do
let!(:chat_team) { create(:chat_team, namespace: group) }
it 'destroys the team too' do
expect_any_instance_of(Mattermost::Team).to receive(:destroy)
destroy_group(group, user, async)
end
end
context 'file system' do
context 'Sidekiq inline' do
before do
# Run sidekiq immediately to check that renamed dir will be removed
perform_enqueued_jobs { destroy_group(group, user, async) }
end
it 'verifies that paths have been deleted' do
expect(gitlab_shell.exists?(project.repository_storage, group.path)).to be_falsey
expect(gitlab_shell.exists?(project.repository_storage, remove_path)).to be_falsey
end
end
end
end
describe 'asynchronous delete' do
it_behaves_like 'group destruction', true
context 'Sidekiq fake' do
before do
# Don't run Sidekiq to verify that group and projects are not actually destroyed
Sidekiq::Testing.fake! { destroy_group(group, user, true) }
end
after do
# Clean up stale directories
gitlab_shell.rm_namespace(project.repository_storage, group.path)
gitlab_shell.rm_namespace(project.repository_storage, remove_path)
end
it 'verifies original paths and projects still exist' do
expect(gitlab_shell.exists?(project.repository_storage, group.path)).to be_truthy
expect(gitlab_shell.exists?(project.repository_storage, remove_path)).to be_falsey
expect(Project.unscoped.count).to eq(1)
expect(Group.unscoped.count).to eq(2)
end
end
context 'potential race conditions' do
context "when the `GroupDestroyWorker` task runs immediately" do
it "deletes the group" do
# Commit the contents of this spec's transaction so far
# so subsequent db connections can see it.
#
# DO NOT REMOVE THIS LINE, even if you see a WARNING with "No
# transaction is currently in progress". Without this, this
# spec will always be green, since the group created in setup
# cannot be seen by any other connections / threads in this spec.
Group.connection.commit_db_transaction
group_record = run_with_new_database_connection do |conn|
conn.execute("SELECT * FROM namespaces WHERE id = #{group.id}").first
end
expect(group_record).not_to be_nil
# Execute the contents of `GroupDestroyWorker` in a separate thread, to
# simulate data manipulation by the Sidekiq worker (different database
# connection / transaction).
expect(GroupDestroyWorker).to receive(:perform_async).and_wrap_original do |m, group_id, user_id|
Thread.new { m[group_id, user_id] }.join(5)
end
# Kick off the initial group destroy in a new thread, so that
# it doesn't share this spec's database transaction.
Thread.new { described_class.new(group, user).async_execute }.join(5)
group_record = run_with_new_database_connection do |conn|
conn.execute("SELECT * FROM namespaces WHERE id = #{group.id}").first
end
expect(group_record).to be_nil
end
end
end
end
describe 'synchronous delete' do
it_behaves_like 'group destruction', false
end
context 'projects in pending_delete' do
before do
project.pending_delete = true
project.save
end
it_behaves_like 'group destruction', false
end
context 'repository removal status is taken into account' do
it 'raises exception' do
expect_next_instance_of(::Projects::DestroyService) do |destroy_service|
expect(destroy_service).to receive(:execute).and_return(false)
end
expect { destroy_group(group, user, false) }
.to raise_error(Groups::DestroyService::DestroyError, "Project #{project.id} can't be deleted" )
end
end
describe 'repository removal' do
before do
destroy_group(group, user, false)
end
context 'legacy storage' do
let!(:project) { create(:project, :legacy_storage, :empty_repo, namespace: group) }
it 'removes repository' do
expect(gitlab_shell.exists?(project.repository_storage, "#{project.disk_path}.git")).to be_falsey
end
end
context 'hashed storage' do
let!(:project) { create(:project, :empty_repo, namespace: group) }
it 'removes repository' do
expect(gitlab_shell.exists?(project.repository_storage, "#{project.disk_path}.git")).to be_falsey
end
end
end
end