Module: Gitlab::Database::Migrations::BatchedBackgroundMigrationHelpers
- Included in:
- Gitlab::Database::MigrationHelpers
- Defined in:
- lib/gitlab/database/migrations/batched_background_migration_helpers.rb
Overview
BatchedBackgroundMigrations are a new approach to scheduling and executing background migrations, which uses persistent state in the database to track each migration. This avoids having to batch over an entire table and schedule a large number of sidekiq jobs upfront. It also provides for more flexibility as the migration runs, as it can be paused and restarted, and have configuration values like the batch size updated dynamically as the migration runs.
For now, these migrations are not considered ready for general use, for more information see the tracking epic: gitlab.com/groups/gitlab-org/-/epics/6751
Constant Summary collapse
- NonExistentMigrationError =
Class.new(StandardError)
- BATCH_SIZE =
Number of rows to process per job
1_000
- SUB_BATCH_SIZE =
Number of rows to process per sub-batch
100
- BATCH_CLASS_NAME =
Default batch class for batched migrations
'PrimaryKeyBatchingStrategy'
- BATCH_MIN_VALUE =
Default minimum value for batched migrations
1
- BATCH_MIN_DELAY =
Minimum delay between batched migrations
2.minutes.freeze
Instance Method Summary collapse
-
#delete_batched_background_migration(job_class_name, table_name, column_name, job_arguments) ⇒ Object
Deletes batched background migration for the given configuration.
- #ensure_batched_background_migration_is_finished(job_class_name:, table_name:, column_name:, job_arguments:, finalize: true) ⇒ Object
- #finalize_batched_background_migration(job_class_name:, table_name:, column_name:, job_arguments:) ⇒ Object
- #gitlab_schema_from_context ⇒ Object
-
#queue_batched_background_migration(job_class_name, batch_table_name, batch_column_name, *job_arguments, job_interval:, batch_min_value: BATCH_MIN_VALUE, batch_max_value: nil, batch_class_name: BATCH_CLASS_NAME, batch_size: BATCH_SIZE, max_batch_size: nil, sub_batch_size: SUB_BATCH_SIZE, gitlab_schema: nil) ⇒ Object
Creates a batched background migration for the given table.
Instance Method Details
#delete_batched_background_migration(job_class_name, table_name, column_name, job_arguments) ⇒ Object
Deletes batched background migration for the given configuration.
job_class_name - The background migration job class as a string table_name - The name of the table the migration iterates over column_name - The name of the column the migration will batch over job_arguments - Migration arguments
Example:
delete_batched_background_migration(
'CopyColumnUsingBackgroundMigrationJob',
:events,
:id,
['column1', 'column2'])
182 183 184 185 186 187 188 189 190 191 |
# File 'lib/gitlab/database/migrations/batched_background_migration_helpers.rb', line 182 def delete_batched_background_migration(job_class_name, table_name, column_name, job_arguments) Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas.require_dml_mode! Gitlab::Database::BackgroundMigration::BatchedMigration.reset_column_information Gitlab::Database::BackgroundMigration::BatchedMigration .for_configuration( gitlab_schema_from_context, job_class_name, table_name, column_name, job_arguments ).delete_all end |
#ensure_batched_background_migration_is_finished(job_class_name:, table_name:, column_name:, job_arguments:, finalize: true) ⇒ Object
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 |
# File 'lib/gitlab/database/migrations/batched_background_migration_helpers.rb', line 201 def ensure_batched_background_migration_is_finished(job_class_name:, table_name:, column_name:, job_arguments:, finalize: true) Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas.require_dml_mode! if transaction_open? raise 'The `ensure_batched_background_migration_is_finished` cannot be run inside a transaction. ' \ 'You can disable transactions by calling `disable_ddl_transaction!` in the body of ' \ 'your migration class.' end Gitlab::Database::BackgroundMigration::BatchedMigration.reset_column_information migration = Gitlab::Database::BackgroundMigration::BatchedMigration.find_for_configuration( Gitlab::Database.gitlab_schemas_for_connection(connection), job_class_name, table_name, column_name, job_arguments ) configuration = { job_class_name: job_class_name, table_name: table_name, column_name: column_name, job_arguments: job_arguments } if ENV['DBLAB_ENVIRONMENT'] && migration.nil? raise NonExistentMigrationError, 'called ensure_batched_background_migration_is_finished with non-existent migration name' end return Gitlab::AppLogger.warn "Could not find batched background migration for the given configuration: #{configuration}" if migration.nil? return if migration.finished? finalize_batched_background_migration(job_class_name: job_class_name, table_name: table_name, column_name: column_name, job_arguments: job_arguments) if finalize return if migration.reload.finished? # rubocop:disable Cop/ActiveRecordAssociationReload raise "Expected batched background migration for the given configuration to be marked as 'finished', " \ "but it is '#{migration.status_name}':" \ "\t#{configuration}" \ "\n\n" \ "Finalize it manually by running the following command in a `bash` or `sh` shell:" \ "\n\n" \ "\tsudo gitlab-rake gitlab:background_migrations:finalize[#{job_class_name},#{table_name},#{column_name},'#{job_arguments.to_json.gsub(',', '\,')}']" \ "\n\n" \ "For more information, check the documentation" \ "\n\n" \ "\thttps://docs.gitlab.com/ee/update/background_migrations.html#database-migrations-failing-because-of-batched-background-migration-not-finished" end |
#finalize_batched_background_migration(job_class_name:, table_name:, column_name:, job_arguments:) ⇒ Object
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 |
# File 'lib/gitlab/database/migrations/batched_background_migration_helpers.rb', line 142 def finalize_batched_background_migration(job_class_name:, table_name:, column_name:, job_arguments:) Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas.require_dml_mode! if transaction_open? raise 'The `finalize_batched_background_migration` cannot be run inside a transaction. ' \ 'You can disable transactions by calling `disable_ddl_transaction!` in the body of ' \ 'your migration class.' end Gitlab::Database::BackgroundMigration::BatchedMigration.reset_column_information migration = Gitlab::Database::BackgroundMigration::BatchedMigration.find_for_configuration( gitlab_schema_from_context, job_class_name, table_name, column_name, job_arguments) raise 'Could not find batched background migration' if migration.nil? with_restored_connection_stack do |restored_connection| Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas.with_suppressed do Gitlab::Database::BackgroundMigration::BatchedMigrationRunner.finalize( job_class_name, table_name, column_name, job_arguments, connection: restored_connection) end end end |
#gitlab_schema_from_context ⇒ Object
193 194 195 196 197 198 199 |
# File 'lib/gitlab/database/migrations/batched_background_migration_helpers.rb', line 193 def gitlab_schema_from_context if respond_to?(:allowed_gitlab_schemas) # Gitlab::Database::Migration::V2_0 Array(allowed_gitlab_schemas).first else # Gitlab::Database::Migration::V1_0 :gitlab_main end end |
#queue_batched_background_migration(job_class_name, batch_table_name, batch_column_name, *job_arguments, job_interval:, batch_min_value: BATCH_MIN_VALUE, batch_max_value: nil, batch_class_name: BATCH_CLASS_NAME, batch_size: BATCH_SIZE, max_batch_size: nil, sub_batch_size: SUB_BATCH_SIZE, gitlab_schema: nil) ⇒ Object
Creates a batched background migration for the given table. A batched migration runs one job at a time, computing the bounds of the next batch based on the current migration settings and the previous batch bounds. Each job’s execution status is tracked in the database as the migration runs. The given job class must be present in the Gitlab::BackgroundMigration module, and the batch class (if specified) must be present in the Gitlab::BackgroundMigration::BatchingStrategies module.
If migration with same job_class_name, table_name, column_name, and job_arguments already exists, this helper will log an warning and not create a new one.
job_class_name - The background migration job class as a string batch_table_name - The name of the table the migration will batch over batch_column_name - The name of the column the migration will batch over job_arguments - Extra arguments to pass to the job instance when the migration runs job_interval - The pause interval between each job’s execution, minimum of 2 minutes batch_min_value - The value in the column the batching will begin at batch_max_value - The value in the column the batching will end at, defaults to ‘SELECT MAX(batch_column)` batch_class_name - The name of the class that will be called to find the range of each next batch batch_size - The maximum number of rows per job sub_batch_size - The maximum number of rows processed per “iteration” within the job
*Returns the created BatchedMigration record*
Example:
queue_batched_background_migration(
'CopyColumnUsingBackgroundMigrationJob',
:events,
:id,
job_interval: 2.minutes,
other_job_arguments: ['column1', 'column2'])
Where the the background migration exists:
class Gitlab::BackgroundMigration::CopyColumnUsingBackgroundMigrationJob
def perform(start_id, end_id, batch_table, batch_column, sub_batch_size, *other_args)
# do something
end
end
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 |
# File 'lib/gitlab/database/migrations/batched_background_migration_helpers.rb', line 60 def queue_batched_background_migration( # rubocop:disable Metrics/ParameterLists job_class_name, batch_table_name, batch_column_name, *job_arguments, job_interval:, batch_min_value: BATCH_MIN_VALUE, batch_max_value: nil, batch_class_name: BATCH_CLASS_NAME, batch_size: BATCH_SIZE, max_batch_size: nil, sub_batch_size: SUB_BATCH_SIZE, gitlab_schema: nil ) Gitlab::Database::QueryAnalyzers::RestrictAllowedSchemas.require_dml_mode! gitlab_schema ||= gitlab_schema_from_context Gitlab::Database::BackgroundMigration::BatchedMigration.reset_column_information if Gitlab::Database::BackgroundMigration::BatchedMigration.for_configuration(gitlab_schema, job_class_name, batch_table_name, batch_column_name, job_arguments).exists? Gitlab::AppLogger.warn "Batched background migration not enqueued because it already exists: " \ "job_class_name: #{job_class_name}, table_name: #{batch_table_name}, column_name: #{batch_column_name}, " \ "job_arguments: #{job_arguments.inspect}" return end job_interval = BATCH_MIN_DELAY if job_interval < BATCH_MIN_DELAY batch_max_value ||= connection.select_value(<<~SQL) SELECT MAX(#{connection.quote_column_name(batch_column_name)}) FROM #{connection.quote_table_name(batch_table_name)} SQL status_event = batch_max_value.nil? ? :finish : :execute batch_max_value ||= batch_min_value migration = Gitlab::Database::BackgroundMigration::BatchedMigration.new( job_class_name: job_class_name, table_name: batch_table_name, column_name: batch_column_name, job_arguments: job_arguments, interval: job_interval, min_value: batch_min_value, max_value: batch_max_value, batch_class_name: batch_class_name, batch_size: batch_size, sub_batch_size: sub_batch_size, status_event: status_event ) if migration.job_class.respond_to?(:job_arguments_count) && migration.job_class.job_arguments_count != job_arguments.count raise "Wrong number of job arguments for #{migration.job_class_name} " \ "(given #{job_arguments.count}, expected #{migration.job_class.job_arguments_count})" end # Below `BatchedMigration` attributes were introduced after the # initial `batched_background_migrations` table was created, so any # migrations that ran relying on initial table schema would not know # about columns introduced later on because this model is not # isolated in migrations, which is why we need to check for existence # of these columns first. if migration.respond_to?(:max_batch_size) migration.max_batch_size = max_batch_size end if migration.respond_to?(:total_tuple_count) # We keep track of the estimated number of tuples to reason later # about the overall progress of a migration. migration.total_tuple_count = Gitlab::Database::SharedModel.using_connection(connection) do Gitlab::Database::PgClass.for_table(batch_table_name)&.cardinality_estimate end end if migration.respond_to?(:gitlab_schema) migration.gitlab_schema = gitlab_schema end migration.save! migration end |