diff options
author | Yorick Peterse <yorickpeterse@gmail.com> | 2016-11-24 15:07:44 +0100 |
---|---|---|
committer | Yorick Peterse <yorickpeterse@gmail.com> | 2016-12-01 13:36:06 +0100 |
commit | 6b4d33566f5f434cc86381a4a1347e42bbe348ee (patch) | |
tree | 441ea1ec210bbef063ec82d470b8e86d38ffec8c /spec/migrations | |
parent | e91afc0dc071f2cb2dde54b12c04bb90d2c65f7b (diff) | |
download | gitlab-ce-process-commit-worker-improvements.tar.gz |
Pass commit data to ProcessCommitWorkerprocess-commit-worker-improvements
By passing commit data to this worker we remove the need for querying
the Git repository for every job. This in turn reduces the time spent
processing each job.
The migration included migrates jobs from the old format to the new
format. For this to work properly it requires downtime as otherwise
workers may start producing errors until they're using a newer version
of the worker code.
Diffstat (limited to 'spec/migrations')
-rw-r--r-- | spec/migrations/migrate_process_commit_worker_jobs_spec.rb | 194 |
1 files changed, 194 insertions, 0 deletions
diff --git a/spec/migrations/migrate_process_commit_worker_jobs_spec.rb b/spec/migrations/migrate_process_commit_worker_jobs_spec.rb new file mode 100644 index 00000000000..52428547a9f --- /dev/null +++ b/spec/migrations/migrate_process_commit_worker_jobs_spec.rb @@ -0,0 +1,194 @@ +require 'spec_helper' +require Rails.root.join('db', 'migrate', '20161124141322_migrate_process_commit_worker_jobs.rb') + +describe MigrateProcessCommitWorkerJobs do + let(:project) { create(:project) } + let(:user) { create(:user) } + let(:commit) { project.commit.raw.raw_commit } + + describe 'Project' do + describe 'find_including_path' do + it 'returns Project instances' do + expect(described_class::Project.find_including_path(project.id)). + to be_an_instance_of(described_class::Project) + end + + it 'selects the full path for every Project' do + migration_project = described_class::Project. + find_including_path(project.id) + + expect(migration_project[:path_with_namespace]). + to eq(project.path_with_namespace) + end + end + + describe '#repository_storage_path' do + it 'returns the storage path for the repository' do + migration_project = described_class::Project. + find_including_path(project.id) + + expect(File.directory?(migration_project.repository_storage_path)). + to eq(true) + end + end + + describe '#repository_path' do + it 'returns the path to the repository' do + migration_project = described_class::Project. + find_including_path(project.id) + + expect(File.directory?(migration_project.repository_path)).to eq(true) + end + end + + describe '#repository' do + it 'returns a Rugged::Repository' do + migration_project = described_class::Project. + find_including_path(project.id) + + expect(migration_project.repository). + to be_an_instance_of(Rugged::Repository) + end + end + end + + describe '#up', :redis do + let(:migration) { described_class.new } + + def job_count + Sidekiq.redis { |r| r.llen('queue:process_commit') } + end + + before do + Sidekiq.redis do |redis| + job = JSON.dump(args: [project.id, user.id, commit.oid]) + redis.lpush('queue:process_commit', job) + end + end + + it 'skips jobs using a project that no longer exists' do + allow(described_class::Project).to receive(:find_including_path). + with(project.id). + and_return(nil) + + migration.up + + expect(job_count).to eq(0) + end + + it 'skips jobs using commits that no longer exist' do + allow_any_instance_of(Rugged::Repository).to receive(:lookup). + with(commit.oid). + and_raise(Rugged::OdbError) + + migration.up + + expect(job_count).to eq(0) + end + + it 'inserts migrated jobs back into the queue' do + migration.up + + expect(job_count).to eq(1) + end + + context 'a migrated job' do + let(:job) do + migration.up + + JSON.load(Sidekiq.redis { |r| r.lpop('queue:process_commit') }) + end + + let(:commit_hash) do + job['args'][2] + end + + it 'includes the project ID' do + expect(job['args'][0]).to eq(project.id) + end + + it 'includes the user ID' do + expect(job['args'][1]).to eq(user.id) + end + + it 'includes the commit ID' do + expect(commit_hash['id']).to eq(commit.oid) + end + + it 'includes the commit message' do + expect(commit_hash['message']).to eq(commit.message) + end + + it 'includes the parent IDs' do + expect(commit_hash['parent_ids']).to eq(commit.parent_ids) + end + + it 'includes the author date' do + expect(commit_hash['authored_date']).to eq(commit.author[:time].to_s) + end + + it 'includes the author name' do + expect(commit_hash['author_name']).to eq(commit.author[:name]) + end + + it 'includes the author Email' do + expect(commit_hash['author_email']).to eq(commit.author[:email]) + end + + it 'includes the commit date' do + expect(commit_hash['committed_date']).to eq(commit.committer[:time].to_s) + end + + it 'includes the committer name' do + expect(commit_hash['committer_name']).to eq(commit.committer[:name]) + end + + it 'includes the committer Email' do + expect(commit_hash['committer_email']).to eq(commit.committer[:email]) + end + end + end + + describe '#down', :redis do + let(:migration) { described_class.new } + + def job_count + Sidekiq.redis { |r| r.llen('queue:process_commit') } + end + + before do + Sidekiq.redis do |redis| + job = JSON.dump(args: [project.id, user.id, commit.oid]) + redis.lpush('queue:process_commit', job) + + migration.up + end + end + + it 'pushes migrated jobs back into the queue' do + migration.down + + expect(job_count).to eq(1) + end + + context 'a migrated job' do + let(:job) do + migration.down + + JSON.load(Sidekiq.redis { |r| r.lpop('queue:process_commit') }) + end + + it 'includes the project ID' do + expect(job['args'][0]).to eq(project.id) + end + + it 'includes the user ID' do + expect(job['args'][1]).to eq(user.id) + end + + it 'includes the commit SHA' do + expect(job['args'][2]).to eq(commit.oid) + end + end + end +end |