1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
|
# frozen_string_literal: true
RSpec.shared_examples 'it runs background migration jobs' do |tracking_database|
describe 'defining the job attributes' do
it 'defines the data_consistency as always' do
expect(described_class.get_data_consistency).to eq(:always)
end
it 'defines the retry count in sidekiq_options' do
expect(described_class.sidekiq_options['retry']).to eq(3)
end
it 'defines the feature_category as database' do
expect(described_class.get_feature_category).to eq(:database)
end
it 'defines the urgency as throttled' do
expect(described_class.get_urgency).to eq(:throttled)
end
it 'defines the loggable_arguments' do
expect(described_class.loggable_arguments).to match_array([0, 1])
end
end
describe '.tracking_database' do
it 'does not raise an error' do
expect { described_class.tracking_database }.not_to raise_error
end
it 'overrides the method to return the tracking database' do
expect(described_class.tracking_database).to eq(tracking_database)
end
end
describe '.minimum_interval' do
it 'returns 2 minutes' do
expect(described_class.minimum_interval).to eq(2.minutes.to_i)
end
end
describe '#perform' do
let(:worker) { described_class.new }
context 'when execute_background_migrations feature flag is disabled' do
before do
stub_feature_flags(execute_background_migrations: false)
end
it 'does not perform the job, reschedules it in the future, and logs a message' do
expect(worker).not_to receive(:perform_with_connection)
expect(Sidekiq.logger).to receive(:info) do |payload|
expect(payload[:class]).to eq(described_class.name)
expect(payload[:database]).to eq(tracking_database)
expect(payload[:message]).to match(/skipping execution, migration rescheduled/)
end
lease_attempts = 3
delay = described_class::BACKGROUND_MIGRATIONS_DELAY
job_args = [10, 20]
freeze_time do
worker.perform('Foo', job_args, lease_attempts)
job = described_class.jobs.find { |job| job['args'] == ['Foo', job_args, lease_attempts] }
expect(job).to be, "Expected the job to be rescheduled with (#{job_args}, #{lease_attempts}), but it was not."
expected_time = delay.to_i + Time.now.to_i
expect(job['at']).to eq(expected_time),
"Expected the job to be rescheduled in #{expected_time} seconds, " \
"but it was rescheduled in #{job['at']} seconds."
end
end
end
context 'when execute_background_migrations feature flag is enabled' do
before do
stub_feature_flags(execute_background_migrations: true)
allow(worker).to receive(:jid).and_return(1)
allow(worker).to receive(:always_perform?).and_return(false)
allow(Postgresql::ReplicationSlot).to receive(:lag_too_great?).and_return(false)
end
it 'performs jobs using the coordinator for the worker' do
expect_next_instance_of(Gitlab::BackgroundMigration::JobCoordinator) do |coordinator|
allow(coordinator).to receive(:with_shared_connection).and_yield
expect(coordinator.worker_class).to eq(described_class)
expect(coordinator).to receive(:perform).with('Foo', [10, 20])
end
worker.perform('Foo', [10, 20])
end
context 'when lease can be obtained' do
let(:coordinator) { double('job coordinator') }
before do
allow(Gitlab::BackgroundMigration).to receive(:coordinator_for_database)
.with(tracking_database)
.and_return(coordinator)
allow(coordinator).to receive(:with_shared_connection).and_yield
end
it 'sets up the shared connection before checking replication' do
expect(coordinator).to receive(:with_shared_connection).and_yield.ordered
expect(Postgresql::ReplicationSlot).to receive(:lag_too_great?).and_return(false).ordered
expect(coordinator).to receive(:perform).with('Foo', [10, 20])
worker.perform('Foo', [10, 20])
end
it 'performs a background migration' do
expect(coordinator).to receive(:perform).with('Foo', [10, 20])
worker.perform('Foo', [10, 20])
end
context 'when lease_attempts is 1' do
it 'performs a background migration' do
expect(coordinator).to receive(:perform).with('Foo', [10, 20])
worker.perform('Foo', [10, 20], 1)
end
end
it 'can run scheduled job and retried job concurrently' do
expect(coordinator)
.to receive(:perform)
.with('Foo', [10, 20])
.exactly(2).time
worker.perform('Foo', [10, 20])
worker.perform('Foo', [10, 20], described_class::MAX_LEASE_ATTEMPTS - 1)
end
it 'sets the class that will be executed as the caller_id' do
expect(coordinator).to receive(:perform) do
expect(Gitlab::ApplicationContext.current).to include('meta.caller_id' => 'Foo')
end
worker.perform('Foo', [10, 20])
end
end
context 'when lease not obtained (migration of same class was performed recently)' do
let(:timeout) { described_class.minimum_interval }
let(:lease_key) { "#{described_class.name}:Foo" }
let(:coordinator) { double('job coordinator') }
before do
allow(Gitlab::BackgroundMigration).to receive(:coordinator_for_database)
.with(tracking_database)
.and_return(coordinator)
allow(coordinator).to receive(:with_shared_connection).and_yield
expect(coordinator).not_to receive(:perform)
Gitlab::ExclusiveLease.new(lease_key, timeout: timeout).try_obtain
end
it 'reschedules the migration and decrements the lease_attempts' do
expect(described_class)
.to receive(:perform_in)
.with(a_kind_of(Numeric), 'Foo', [10, 20], 4)
worker.perform('Foo', [10, 20], 5)
end
context 'when lease_attempts is 1' do
let(:lease_key) { "#{described_class.name}:Foo:retried" }
it 'reschedules the migration and decrements the lease_attempts' do
expect(described_class)
.to receive(:perform_in)
.with(a_kind_of(Numeric), 'Foo', [10, 20], 0)
worker.perform('Foo', [10, 20], 1)
end
end
context 'when lease_attempts is 0' do
let(:lease_key) { "#{described_class.name}:Foo:retried" }
it 'gives up performing the migration' do
expect(described_class).not_to receive(:perform_in)
expect(Sidekiq.logger).to receive(:warn).with(
class: 'Foo',
message: 'Job could not get an exclusive lease after several tries. Giving up.',
job_id: 1)
worker.perform('Foo', [10, 20], 0)
end
end
end
context 'when database is not healthy' do
before do
expect(Postgresql::ReplicationSlot).to receive(:lag_too_great?).and_return(true)
end
it 'reschedules a migration if the database is not healthy' do
expect(described_class)
.to receive(:perform_in)
.with(a_kind_of(Numeric), 'Foo', [10, 20], 4)
worker.perform('Foo', [10, 20])
end
it 'increments the unhealthy counter' do
counter = Gitlab::Metrics.counter(:background_migration_database_health_reschedules, 'msg')
expect(described_class).to receive(:perform_in)
expect { worker.perform('Foo', [10, 20]) }.to change { counter.get(db_config_name: tracking_database) }.by(1)
end
context 'when lease_attempts is 0' do
it 'gives up performing the migration' do
expect(described_class).not_to receive(:perform_in)
expect(Sidekiq.logger).to receive(:warn).with(
class: 'Foo',
message: 'Database was unhealthy after several tries. Giving up.',
job_id: 1)
worker.perform('Foo', [10, 20], 0)
end
end
end
end
end
end
|