diff options
Diffstat (limited to 'spec/lib/gitlab/database/batch_count_spec.rb')
-rw-r--r-- | spec/lib/gitlab/database/batch_count_spec.rb | 177 |
1 files changed, 123 insertions, 54 deletions
diff --git a/spec/lib/gitlab/database/batch_count_spec.rb b/spec/lib/gitlab/database/batch_count_spec.rb index 811d4fad95c..a87b0c1a3a8 100644 --- a/spec/lib/gitlab/database/batch_count_spec.rb +++ b/spec/lib/gitlab/database/batch_count_spec.rb @@ -86,48 +86,48 @@ RSpec.describe Gitlab::Database::BatchCount do query: batch_count_query, message: 'Query has been canceled with message: query timed out' ) - expect(subject.call(model, column, batch_size: batch_size, start: 0)).to eq(-1) + expect(subject.call(model, column, batch_size: batch_size, start: 0)).to eq(fallback) end end end describe '#batch_count' do it 'counts table' do - expect(described_class.batch_count(model)).to eq(5) + expect(described_class.batch_count(model)).to eq(model.count) end it 'counts with :id field' do - expect(described_class.batch_count(model, :id)).to eq(5) + expect(described_class.batch_count(model, :id)).to eq(model.count) end it 'counts with "id" field' do - expect(described_class.batch_count(model, 'id')).to eq(5) + expect(described_class.batch_count(model, 'id')).to eq(model.count) end it 'counts with table.id field' do - expect(described_class.batch_count(model, "#{model.table_name}.id")).to eq(5) + expect(described_class.batch_count(model, "#{model.table_name}.id")).to eq(model.count) end it 'counts with Arel column' do - expect(described_class.batch_count(model, model.arel_table[:id])).to eq(5) + expect(described_class.batch_count(model, model.arel_table[:id])).to eq(model.count) end it 'counts table with batch_size 50K' do - expect(described_class.batch_count(model, batch_size: 50_000)).to eq(5) + expect(described_class.batch_count(model, batch_size: 50_000)).to eq(model.count) end it 'will not count table with a batch size less than allowed' do expect(described_class.batch_count(model, batch_size: small_batch_size)).to eq(fallback) end - it 'counts with a small edge case batch_sizes than result' do + it 'produces the same result with different batch sizes' do stub_const('Gitlab::Database::BatchCounter::MIN_REQUIRED_BATCH_SIZE', 0) - [1, 2, 4, 5, 6].each { |i| expect(described_class.batch_count(model, batch_size: i)).to eq(5) } + [1, 2, 4, 5, 6].each { |i| expect(described_class.batch_count(model, batch_size: i)).to eq(model.count) } end it 'counts with a start and finish' do - expect(described_class.batch_count(model, start: model.minimum(:id), finish: model.maximum(:id))).to eq(5) + expect(described_class.batch_count(model, start: model.minimum(:id), finish: model.maximum(:id))).to eq(model.count) end it 'stops counting when finish value is reached' do @@ -217,6 +217,113 @@ RSpec.describe Gitlab::Database::BatchCount do end end + describe '#batch_count_with_timeout' do + it 'counts table' do + expect(described_class.batch_count_with_timeout(model)).to eq({ status: :completed, count: model.count }) + end + + it 'counts with :id field' do + expect(described_class.batch_count_with_timeout(model, :id)).to eq({ status: :completed, count: model.count }) + end + + it 'counts with "id" field' do + expect(described_class.batch_count_with_timeout(model, 'id')).to eq({ status: :completed, count: model.count }) + end + + it 'counts with table.id field' do + expect(described_class.batch_count_with_timeout(model, "#{model.table_name}.id")).to eq({ status: :completed, count: model.count }) + end + + it 'counts with Arel column' do + expect(described_class.batch_count_with_timeout(model, model.arel_table[:id])).to eq({ status: :completed, count: model.count }) + end + + it 'counts table with batch_size 50K' do + expect(described_class.batch_count_with_timeout(model, batch_size: 50_000)).to eq({ status: :completed, count: model.count }) + end + + it 'will not count table with a batch size less than allowed' do + expect(described_class.batch_count_with_timeout(model, batch_size: small_batch_size)).to eq({ status: :bad_config }) + end + + it 'produces the same result with different batch sizes' do + stub_const('Gitlab::Database::BatchCounter::MIN_REQUIRED_BATCH_SIZE', 0) + + [1, 2, 4, 5, 6].each { |i| expect(described_class.batch_count_with_timeout(model, batch_size: i)).to eq({ status: :completed, count: model.count }) } + end + + it 'counts with a start and finish' do + expect(described_class.batch_count_with_timeout(model, start: model.minimum(:id), finish: model.maximum(:id))).to eq({ status: :completed, count: model.count }) + end + + it 'stops counting when finish value is reached' do + stub_const('Gitlab::Database::BatchCounter::MIN_REQUIRED_BATCH_SIZE', 0) + + expect(described_class.batch_count_with_timeout(model, + start: model.minimum(:id), + finish: model.maximum(:id) - 1, # Do not count the last record + batch_size: model.count - 2 # Ensure there are multiple batches + )).to eq({ status: :completed, count: model.count - 1 }) + end + + it 'returns a partial count when timeout elapses' do + stub_const('Gitlab::Database::BatchCounter::MIN_REQUIRED_BATCH_SIZE', 0) + + expect(::Gitlab::Metrics::System).to receive(:monotonic_time).and_return(1, 10, 300) + + expect( + described_class.batch_count_with_timeout(model, batch_size: 1, timeout: 250.seconds) + ).to eq({ status: :timeout, partial_results: 1, continue_from: model.minimum(:id) + 1 }) + end + + it 'starts counting from a given partial result' do + expect(described_class.batch_count_with_timeout(model, partial_results: 3)).to eq({ status: :completed, count: 3 + model.count }) + end + + it_behaves_like 'when a transaction is open' do + subject { described_class.batch_count_with_timeout(model) } + end + + it_behaves_like 'when batch fetch query is canceled' do + let(:mode) { :itself } + let(:operation) { :count } + let(:operation_args) { nil } + let(:column) { nil } + let(:fallback) { { status: :cancelled } } + + subject { described_class.method(:batch_count_with_timeout) } + end + + context 'disallowed_configurations' do + include_examples 'disallowed configurations', :batch_count do + let(:args) { [Issue] } + let(:default_batch_size) { Gitlab::Database::BatchCounter::DEFAULT_BATCH_SIZE } + end + + it 'raises an error if distinct count is requested' do + expect { described_class.batch_count_with_timeout(model.distinct(column)) }.to raise_error 'Use distinct count for optimized distinct counting' + end + end + + context 'when a relation is grouped' do + let!(:one_more_issue) { create(:issue, author: user, project: model.first.project) } + + before do + stub_const('Gitlab::Database::BatchCounter::MIN_REQUIRED_BATCH_SIZE', 1) + end + + context 'count by default column' do + let(:count) do + described_class.batch_count_with_timeout(model.group(column), batch_size: 2) + end + + it 'counts grouped records' do + expect(count).to eq({ status: :completed, count: { user.id => 4, another_user.id => 2 } }) + end + end + end + end + describe '#batch_distinct_count' do it 'counts with column field' do expect(described_class.batch_distinct_count(model, column)).to eq(2) @@ -242,7 +349,7 @@ RSpec.describe Gitlab::Database::BatchCount do expect(described_class.batch_distinct_count(model, column, batch_size: small_batch_size)).to eq(fallback) end - it 'counts with a small edge case batch_sizes than result' do + it 'produces the same result with different batch sizes' do stub_const('Gitlab::Database::BatchCounter::MIN_REQUIRED_BATCH_SIZE', 0) [1, 2, 4, 5, 6].each { |i| expect(described_class.batch_distinct_count(model, column, batch_size: i)).to eq(2) } @@ -386,56 +493,18 @@ RSpec.describe Gitlab::Database::BatchCount do end describe '#batch_average' do - let(:model) { Issue } let(:column) { :weight } before do - Issue.update_all(weight: 2) - end - - it 'returns the average of values in the given column' do - expect(described_class.batch_average(model, column)).to eq(2) - end - - it 'works when given an Arel column' do - expect(described_class.batch_average(model, model.arel_table[column])).to eq(2) - end - - it 'works with a batch size of 50K' do - expect(described_class.batch_average(model, column, batch_size: 50_000)).to eq(2) - end - - it 'works with start and finish provided' do - expect(described_class.batch_average(model, column, start: model.minimum(:id), finish: model.maximum(:id))).to eq(2) + allow_next_instance_of(Gitlab::Database::BatchAverageCounter) do |instance| + allow(instance).to receive(:count).and_return + end end - it "defaults the batch size to #{Gitlab::Database::BatchCounter::DEFAULT_AVERAGE_BATCH_SIZE}" do - min_id = model.minimum(:id) - relation = instance_double(ActiveRecord::Relation) - allow(model).to receive_message_chain(:select, public_send: relation) - batch_end_id = min_id + calculate_batch_size(Gitlab::Database::BatchCounter::DEFAULT_AVERAGE_BATCH_SIZE) - - expect(relation).to receive(:where).with("id" => min_id..batch_end_id).and_return(double(send: 1)) + it 'calls BatchAverageCounter' do + expect(Gitlab::Database::BatchAverageCounter).to receive(:new).with(model, column).and_call_original described_class.batch_average(model, column) end - - it_behaves_like 'when a transaction is open' do - subject { described_class.batch_average(model, column) } - end - - it_behaves_like 'disallowed configurations', :batch_average do - let(:args) { [model, column] } - let(:default_batch_size) { Gitlab::Database::BatchCounter::DEFAULT_AVERAGE_BATCH_SIZE } - let(:small_batch_size) { Gitlab::Database::BatchCounter::DEFAULT_AVERAGE_BATCH_SIZE - 1 } - end - - it_behaves_like 'when batch fetch query is canceled' do - let(:mode) { :itself } - let(:operation) { :average } - let(:operation_args) { [column] } - - subject { described_class.method(:batch_average) } - end end end |