gitlab-org--gitlab-foss/spec/lib/gitlab/cache/ci/project_pipeline_status_spe...

305 lines
10 KiB
Ruby
Raw Normal View History

require 'spec_helper'
describe Gitlab::Cache::Ci::ProjectPipelineStatus, :redis do
Use CTEs for nested groups and authorizations This commit introduces the usage of Common Table Expressions (CTEs) to efficiently retrieve nested group hierarchies, without having to rely on the "routes" table (which is an _incredibly_ inefficient way of getting the data). This requires a patch to ActiveRecord (found in the added initializer) to work properly as ActiveRecord doesn't support WITH statements properly out of the box. Unfortunately MySQL provides no efficient way of getting nested groups. For example, the old routes setup could easily take 5-10 seconds depending on the amount of "routes" in a database. Providing vastly different logic for both MySQL and PostgreSQL will negatively impact the development process. Because of this the various nested groups related methods return empty relations when used in combination with MySQL. For project authorizations the logic is split up into two classes: * Gitlab::ProjectAuthorizations::WithNestedGroups * Gitlab::ProjectAuthorizations::WithoutNestedGroups Both classes get the fresh project authorizations (= as they should be in the "project_authorizations" table), including nested groups if PostgreSQL is used. The logic of these two classes is quite different apart from their public interface. This complicates development a bit, but unfortunately there is no way around this. This commit also introduces Gitlab::GroupHierarchy. This class can be used to get the ancestors and descendants of a base relation, or both by using a UNION. This in turn is used by methods such as: * Namespace#ancestors * Namespace#descendants * User#all_expanded_groups Again this class relies on CTEs and thus only works on PostgreSQL. The Namespace methods will return an empty relation when MySQL is used, while User#all_expanded_groups will return only the groups a user is a direct member of. Performance wise the impact is quite large. For example, on GitLab.com Namespace#descendants used to take around 580 ms to retrieve data for a particular user. Using CTEs we are able to reduce this down to roughly 1 millisecond, returning the exact same data. == On The Fly Refreshing Refreshing of authorizations on the fly (= when users.authorized_projects_populated was not set) is removed with this commit. This simplifies the code, and ensures any queries used for authorizations are not mutated because they are executed in a Rails scope (e.g. Project.visible_to_user). This commit includes a migration to schedule refreshing authorizations for all users, ensuring all of them have their authorizations in place. Said migration schedules users in batches of 5000, with 5 minutes between every batch to smear the load around a bit. == Spec Changes This commit also introduces some changes to various specs. For example, some specs for ProjectTeam assumed that creating a personal project would _not_ lead to the owner having access, which is incorrect. Because we also no longer refresh authorizations on the fly for new users some code had to be added to the "empty_project" factory. This chunk of code ensures that the owner's permissions are refreshed after creating the project, something that is normally done in Projects::CreateService.
2017-04-24 15:19:22 +00:00
let!(:project) { create(:project) }
let(:pipeline_status) { described_class.new(project) }
let(:cache_key) { "projects/#{project.id}/pipeline_status" }
describe '.load_for_project' do
it "loads the status" do
expect_any_instance_of(described_class).to receive(:load_status)
described_class.load_for_project(project)
end
end
describe 'loading in batches' do
let(:status) { 'success' }
let(:sha) { '424d1b73bc0d3cb726eb7dc4ce17a4d48552f8c6' }
let(:ref) { 'master' }
let(:pipeline_info) { { sha: sha, status: status, ref: ref } }
Use CTEs for nested groups and authorizations This commit introduces the usage of Common Table Expressions (CTEs) to efficiently retrieve nested group hierarchies, without having to rely on the "routes" table (which is an _incredibly_ inefficient way of getting the data). This requires a patch to ActiveRecord (found in the added initializer) to work properly as ActiveRecord doesn't support WITH statements properly out of the box. Unfortunately MySQL provides no efficient way of getting nested groups. For example, the old routes setup could easily take 5-10 seconds depending on the amount of "routes" in a database. Providing vastly different logic for both MySQL and PostgreSQL will negatively impact the development process. Because of this the various nested groups related methods return empty relations when used in combination with MySQL. For project authorizations the logic is split up into two classes: * Gitlab::ProjectAuthorizations::WithNestedGroups * Gitlab::ProjectAuthorizations::WithoutNestedGroups Both classes get the fresh project authorizations (= as they should be in the "project_authorizations" table), including nested groups if PostgreSQL is used. The logic of these two classes is quite different apart from their public interface. This complicates development a bit, but unfortunately there is no way around this. This commit also introduces Gitlab::GroupHierarchy. This class can be used to get the ancestors and descendants of a base relation, or both by using a UNION. This in turn is used by methods such as: * Namespace#ancestors * Namespace#descendants * User#all_expanded_groups Again this class relies on CTEs and thus only works on PostgreSQL. The Namespace methods will return an empty relation when MySQL is used, while User#all_expanded_groups will return only the groups a user is a direct member of. Performance wise the impact is quite large. For example, on GitLab.com Namespace#descendants used to take around 580 ms to retrieve data for a particular user. Using CTEs we are able to reduce this down to roughly 1 millisecond, returning the exact same data. == On The Fly Refreshing Refreshing of authorizations on the fly (= when users.authorized_projects_populated was not set) is removed with this commit. This simplifies the code, and ensures any queries used for authorizations are not mutated because they are executed in a Rails scope (e.g. Project.visible_to_user). This commit includes a migration to schedule refreshing authorizations for all users, ensuring all of them have their authorizations in place. Said migration schedules users in batches of 5000, with 5 minutes between every batch to smear the load around a bit. == Spec Changes This commit also introduces some changes to various specs. For example, some specs for ProjectTeam assumed that creating a personal project would _not_ lead to the owner having access, which is incorrect. Because we also no longer refresh authorizations on the fly for new users some code had to be added to the "empty_project" factory. This chunk of code ensures that the owner's permissions are refreshed after creating the project, something that is normally done in Projects::CreateService.
2017-04-24 15:19:22 +00:00
let!(:project_without_status) { create(:project) }
describe '.load_in_batch_for_projects' do
it 'preloads pipeline_status on projects' do
described_class.load_in_batch_for_projects([project])
# Don't call the accessor that would lazy load the variable
expect(project.instance_variable_get('@pipeline_status')).to be_a(described_class)
end
describe 'without a status in redis' do
it 'loads the status from a commit when it was not in redis' do
empty_status = { sha: nil, status: nil, ref: nil }
fake_pipeline = described_class.new(
project_without_status,
pipeline_info: empty_status,
loaded_from_cache: false
)
2017-06-21 13:48:12 +00:00
expect(described_class).to receive(:new)
.with(project_without_status,
pipeline_info: empty_status,
2017-06-21 13:48:12 +00:00
loaded_from_cache: false)
.and_return(fake_pipeline)
expect(fake_pipeline).to receive(:load_from_project)
expect(fake_pipeline).to receive(:store_in_cache)
described_class.load_in_batch_for_projects([project_without_status])
end
it 'only connects to redis twice' do
# Once to load, once to store in the cache
expect(Gitlab::Redis).to receive(:with).exactly(2).and_call_original
described_class.load_in_batch_for_projects([project_without_status])
expect(project_without_status.pipeline_status).not_to be_nil
end
end
describe 'when a status was cached in redis' do
before do
Gitlab::Redis.with do |redis|
redis.mapped_hmset(cache_key,
{ sha: sha, status: status, ref: ref })
end
end
it 'loads the correct status' do
described_class.load_in_batch_for_projects([project])
pipeline_status = project.instance_variable_get('@pipeline_status')
expect(pipeline_status.sha).to eq(sha)
expect(pipeline_status.status).to eq(status)
expect(pipeline_status.ref).to eq(ref)
end
it 'only connects to redis once' do
expect(Gitlab::Redis).to receive(:with).exactly(1).and_call_original
described_class.load_in_batch_for_projects([project])
expect(project.pipeline_status).not_to be_nil
end
it "doesn't load the status separatly" do
expect_any_instance_of(described_class).not_to receive(:load_from_project)
expect_any_instance_of(described_class).not_to receive(:load_from_cache)
described_class.load_in_batch_for_projects([project])
end
end
end
describe '.cached_results_for_projects' do
it 'loads a status from redis for all projects' do
Gitlab::Redis.with do |redis|
redis.mapped_hmset(cache_key, { sha: sha, status: status, ref: ref })
end
result = [{ loaded_from_cache: false, pipeline_info: { sha: nil, status: nil, ref: nil } },
{ loaded_from_cache: true, pipeline_info: pipeline_info }]
expect(described_class.cached_results_for_projects([project_without_status, project])).to eq(result)
end
end
end
describe '.update_for_pipeline' do
it 'refreshes the cache if nescessary' do
pipeline = build_stubbed(:ci_pipeline,
sha: '123456', status: 'success', ref: 'master')
fake_status = double
2017-06-21 13:48:12 +00:00
expect(described_class).to receive(:new)
.with(pipeline.project,
pipeline_info: {
sha: '123456', status: 'success', ref: 'master'
2017-06-21 13:48:12 +00:00
})
.and_return(fake_status)
expect(fake_status).to receive(:store_in_cache_if_needed)
described_class.update_for_pipeline(pipeline)
end
end
describe '#has_status?' do
it "is false when the status wasn't loaded yet" do
expect(pipeline_status.has_status?).to be_falsy
end
it 'is true when all status information was loaded' do
fake_commit = double
allow(fake_commit).to receive(:status).and_return('failed')
allow(fake_commit).to receive(:sha).and_return('failed424d1b73bc0d3cb726eb7dc4ce17a4d48552f8c6')
allow(pipeline_status).to receive(:commit).and_return(fake_commit)
allow(pipeline_status).to receive(:has_cache?).and_return(false)
pipeline_status.load_status
expect(pipeline_status.has_status?).to be_truthy
end
end
describe '#load_status' do
it 'loads the status from the cache when there is one' do
expect(pipeline_status).to receive(:has_cache?).and_return(true)
expect(pipeline_status).to receive(:load_from_cache)
pipeline_status.load_status
end
it 'loads the status from the project commit when there is no cache' do
allow(pipeline_status).to receive(:has_cache?).and_return(false)
expect(pipeline_status).to receive(:load_from_project)
pipeline_status.load_status
end
it 'stores the status in the cache when it loading it from the project' do
allow(pipeline_status).to receive(:has_cache?).and_return(false)
allow(pipeline_status).to receive(:load_from_project)
expect(pipeline_status).to receive(:store_in_cache)
pipeline_status.load_status
end
it 'sets the state to loaded' do
pipeline_status.load_status
expect(pipeline_status).to be_loaded
end
it 'only loads the status once' do
expect(pipeline_status).to receive(:has_cache?).and_return(true).exactly(1)
expect(pipeline_status).to receive(:load_from_cache).exactly(1)
pipeline_status.load_status
pipeline_status.load_status
end
end
describe "#load_from_project" do
let!(:pipeline) { create(:ci_pipeline, :success, project: project, sha: project.commit.sha) }
it 'reads the status from the pipeline for the commit' do
pipeline_status.load_from_project
expect(pipeline_status.status).to eq('success')
expect(pipeline_status.sha).to eq(project.commit.sha)
expect(pipeline_status.ref).to eq(project.default_branch)
end
it "doesn't fail for an empty project" do
status_for_empty_commit = described_class.new(create(:empty_project))
status_for_empty_commit.load_status
expect(status_for_empty_commit).to be_loaded
end
end
describe "#store_in_cache", :redis do
it "sets the object in redis" do
pipeline_status.sha = '123456'
pipeline_status.status = 'failed'
pipeline_status.store_in_cache
read_sha, read_status = Gitlab::Redis.with { |redis| redis.hmget(cache_key, :sha, :status) }
expect(read_sha).to eq('123456')
expect(read_status).to eq('failed')
end
end
describe '#store_in_cache_if_needed', :redis do
it 'stores the state in the cache when the sha is the HEAD of the project' do
create(:ci_pipeline, :success, project: project, sha: project.commit.sha)
pipeline_status = described_class.load_for_project(project)
pipeline_status.store_in_cache_if_needed
sha, status, ref = Gitlab::Redis.with { |redis| redis.hmget(cache_key, :sha, :status, :ref) }
expect(sha).not_to be_nil
expect(status).not_to be_nil
expect(ref).not_to be_nil
end
it "doesn't store the status in redis when the sha is not the head of the project" do
other_status = described_class.new(
project,
pipeline_info: { sha: "123456", status: "failed" }
)
other_status.store_in_cache_if_needed
sha, status = Gitlab::Redis.with { |redis| redis.hmget(cache_key, :sha, :status) }
expect(sha).to be_nil
expect(status).to be_nil
end
it "deletes the cache if the repository doesn't have a head commit" do
empty_project = create(:empty_project)
Gitlab::Redis.with do |redis|
redis.mapped_hmset(cache_key,
{ sha: 'sha', status: 'pending', ref: 'master' })
end
other_status = described_class.new(empty_project,
pipeline_info: {
sha: "123456", status: "failed"
})
other_status.store_in_cache_if_needed
sha, status, ref = Gitlab::Redis.with { |redis| redis.hmget("projects/#{empty_project.id}/pipeline_status", :sha, :status, :ref) }
expect(sha).to be_nil
expect(status).to be_nil
expect(ref).to be_nil
end
end
describe "with a status in redis", :redis do
let(:status) { 'success' }
let(:sha) { '424d1b73bc0d3cb726eb7dc4ce17a4d48552f8c6' }
let(:ref) { 'master' }
before do
Gitlab::Redis.with do |redis|
redis.mapped_hmset(cache_key,
{ sha: sha, status: status, ref: ref })
end
end
describe '#load_from_cache' do
it 'reads the status from redis' do
pipeline_status.load_from_cache
expect(pipeline_status.sha).to eq(sha)
expect(pipeline_status.status).to eq(status)
expect(pipeline_status.ref).to eq(ref)
end
end
describe '#has_cache?' do
it 'knows the status is cached' do
expect(pipeline_status.has_cache?).to be_truthy
end
end
describe '#delete_from_cache' do
it 'deletes values from redis' do
pipeline_status.delete_from_cache
key_exists = Gitlab::Redis.with { |redis| redis.exists(cache_key) }
expect(key_exists).to be_falsy
end
end
end
end