diff options
author | Toon Claes <toon@gitlab.com> | 2018-06-26 15:12:29 +0200 |
---|---|---|
committer | Toon Claes <toon@gitlab.com> | 2018-06-26 21:11:36 +0200 |
commit | dd1a331fec305df16c94d81c018d530d2282a731 (patch) | |
tree | bec01ef088914e126f4aca461f8abfc4fc27e2ae | |
parent | 324c78d6ea24e8e5b0a8556aaeb5fb962485eb5a (diff) | |
download | gitlab-ce-tc-repo-check-per-shard.tar.gz |
Add RepositoryCheck::DispatchWorker to start worker per shardtc-repo-check-per-shard
The RepositoryCheck::DispatchWorker will start a
RepositoryCheck::BatchWorker for each healthy shard.
Closes gitlab-org/gitlab-ce#48042
-rw-r--r-- | app/workers/all_queues.yml | 3 | ||||
-rw-r--r-- | app/workers/repository_check/batch_worker.rb | 17 | ||||
-rw-r--r-- | app/workers/repository_check/dispatch_worker.rb | 15 | ||||
-rw-r--r-- | changelogs/unreleased/tc-repo-check-per-shard.yml | 5 | ||||
-rw-r--r-- | config/initializers/1_settings.rb | 2 | ||||
-rw-r--r-- | spec/workers/repository_check/batch_worker_spec.rb | 22 | ||||
-rw-r--r-- | spec/workers/repository_check/dispatch_worker_spec.rb | 19 |
7 files changed, 72 insertions, 11 deletions
diff --git a/app/workers/all_queues.yml b/app/workers/all_queues.yml index 026f756582d..d06f51b1828 100644 --- a/app/workers/all_queues.yml +++ b/app/workers/all_queues.yml @@ -11,7 +11,7 @@ - cronjob:remove_old_web_hook_logs - cronjob:remove_unreferenced_lfs_objects - cronjob:repository_archive_cache -- cronjob:repository_check_batch +- cronjob:repository_check_dispatch - cronjob:requests_profiles - cronjob:schedule_update_user_activity - cronjob:stuck_ci_jobs @@ -71,6 +71,7 @@ - pipeline_processing:update_head_pipeline_for_merge_request - repository_check:repository_check_clear +- repository_check:repository_check_batch - repository_check:repository_check_single_repository - default diff --git a/app/workers/repository_check/batch_worker.rb b/app/workers/repository_check/batch_worker.rb index 72f0a9b0619..fcc12bfe09d 100644 --- a/app/workers/repository_check/batch_worker.rb +++ b/app/workers/repository_check/batch_worker.rb @@ -1,13 +1,18 @@ module RepositoryCheck class BatchWorker include ApplicationWorker - include CronjobQueue + include RepositoryCheckQueue RUN_TIME = 3600 BATCH_SIZE = 10_000 - def perform + attr_reader :shard_name + + def perform(shard_name) + @shard_name = shard_name + return unless Gitlab::CurrentSettings.repository_checks_enabled + return unless Gitlab::ShardHealthCache.healthy_shard?(shard_name) start = Time.now @@ -37,18 +42,22 @@ module RepositoryCheck end def never_checked_project_ids(batch_size) - Project.where(last_repository_check_at: nil) + projects_on_shard.where(last_repository_check_at: nil) .where('created_at < ?', 24.hours.ago) .limit(batch_size).pluck(:id) end def old_checked_project_ids(batch_size) - Project.where.not(last_repository_check_at: nil) + projects_on_shard.where.not(last_repository_check_at: nil) .where('last_repository_check_at < ?', 1.month.ago) .reorder(last_repository_check_at: :asc) .limit(batch_size).pluck(:id) end + def projects_on_shard + Project.where(repository_storage: shard_name) + end + def try_obtain_lease(id) # Use a 24-hour timeout because on servers/projects where 'git fsck' is # super slow we definitely do not want to run it twice in parallel. diff --git a/app/workers/repository_check/dispatch_worker.rb b/app/workers/repository_check/dispatch_worker.rb new file mode 100644 index 00000000000..403479a8497 --- /dev/null +++ b/app/workers/repository_check/dispatch_worker.rb @@ -0,0 +1,15 @@ +module RepositoryCheck + class DispatchWorker + include ApplicationWorker + include CronjobQueue + include ::EachShardWorker + + def perform + return unless Gitlab::CurrentSettings.repository_checks_enabled + + each_shard do |shard_name| + RepositoryCheck::BatchWorker.perform_async(shard_name) + end + end + end +end diff --git a/changelogs/unreleased/tc-repo-check-per-shard.yml b/changelogs/unreleased/tc-repo-check-per-shard.yml new file mode 100644 index 00000000000..e402b72ba4b --- /dev/null +++ b/changelogs/unreleased/tc-repo-check-per-shard.yml @@ -0,0 +1,5 @@ +--- +title: Run repository checks parallel for each shard +merge_request: 20179 +author: +type: added diff --git a/config/initializers/1_settings.rb b/config/initializers/1_settings.rb index 3d3448cb4d6..3236b2c2f15 100644 --- a/config/initializers/1_settings.rb +++ b/config/initializers/1_settings.rb @@ -279,7 +279,7 @@ Settings.cron_jobs['expire_build_artifacts_worker']['cron'] ||= '50 * * * *' Settings.cron_jobs['expire_build_artifacts_worker']['job_class'] = 'ExpireBuildArtifactsWorker' Settings.cron_jobs['repository_check_worker'] ||= Settingslogic.new({}) Settings.cron_jobs['repository_check_worker']['cron'] ||= '20 * * * *' -Settings.cron_jobs['repository_check_worker']['job_class'] = 'RepositoryCheck::BatchWorker' +Settings.cron_jobs['repository_check_worker']['job_class'] = 'RepositoryCheck::DispatchWorker' Settings.cron_jobs['admin_email_worker'] ||= Settingslogic.new({}) Settings.cron_jobs['admin_email_worker']['cron'] ||= '0 0 * * 0' Settings.cron_jobs['admin_email_worker']['job_class'] = 'AdminEmailWorker' diff --git a/spec/workers/repository_check/batch_worker_spec.rb b/spec/workers/repository_check/batch_worker_spec.rb index 6cd27d2fafb..a070e6b1133 100644 --- a/spec/workers/repository_check/batch_worker_spec.rb +++ b/spec/workers/repository_check/batch_worker_spec.rb @@ -1,14 +1,19 @@ require 'spec_helper' describe RepositoryCheck::BatchWorker do + let(:shard_name) { 'default' } subject { described_class.new } + before do + Gitlab::ShardHealthCache.update([shard_name]) + end + it 'prefers projects that have never been checked' do projects = create_list(:project, 3, created_at: 1.week.ago) projects[0].update_column(:last_repository_check_at, 4.months.ago) projects[2].update_column(:last_repository_check_at, 3.months.ago) - expect(subject.perform).to eq(projects.values_at(1, 0, 2).map(&:id)) + expect(subject.perform(shard_name)).to eq(projects.values_at(1, 0, 2).map(&:id)) end it 'sorts projects by last_repository_check_at' do @@ -17,7 +22,7 @@ describe RepositoryCheck::BatchWorker do projects[1].update_column(:last_repository_check_at, 4.months.ago) projects[2].update_column(:last_repository_check_at, 3.months.ago) - expect(subject.perform).to eq(projects.values_at(1, 2, 0).map(&:id)) + expect(subject.perform(shard_name)).to eq(projects.values_at(1, 2, 0).map(&:id)) end it 'excludes projects that were checked recently' do @@ -26,7 +31,7 @@ describe RepositoryCheck::BatchWorker do projects[1].update_column(:last_repository_check_at, 2.months.ago) projects[2].update_column(:last_repository_check_at, 3.days.ago) - expect(subject.perform).to eq([projects[1].id]) + expect(subject.perform(shard_name)).to eq([projects[1].id]) end it 'does nothing when repository checks are disabled' do @@ -34,13 +39,20 @@ describe RepositoryCheck::BatchWorker do stub_application_setting(repository_checks_enabled: false) - expect(subject.perform).to eq(nil) + expect(subject.perform(shard_name)).to eq(nil) + end + + it 'does nothing when shard is unhealthy' do + shard_name = 'broken' + create(:project, created_at: 1.week.ago, repository_storage: shard_name) + + expect(subject.perform(shard_name)).to eq(nil) end it 'skips projects created less than 24 hours ago' do project = create(:project) project.update_column(:created_at, 23.hours.ago) - expect(subject.perform).to eq([]) + expect(subject.perform(shard_name)).to eq([]) end end diff --git a/spec/workers/repository_check/dispatch_worker_spec.rb b/spec/workers/repository_check/dispatch_worker_spec.rb new file mode 100644 index 00000000000..111096a4513 --- /dev/null +++ b/spec/workers/repository_check/dispatch_worker_spec.rb @@ -0,0 +1,19 @@ +require 'spec_helper' + +describe RepositoryCheck::DispatchWorker do + subject { described_class.new } + + it 'does nothing when repository checks are disabled' do + stub_application_setting(repository_checks_enabled: false) + + expect(RepositoryCheck::BatchWorker).not_to receive(:perform_async) + + subject.perform + end + + it 'dispatches work to RepositoryCheck::BatchWorker' do + expect(RepositoryCheck::BatchWorker).to receive(:perform_async) + + subject.perform + end +end |