if command not in ('populate', 'migrate'):
raise Exception(
'Command must either be "populate" or "migrate" but was "%s"' %
command)
if command == 'migrate' and \
not any((args.use_mysql, args.use_postgresql, args.use_oracle)):
raise Exception(
'At least one destination database must be chosen with '
'--use-<database type>')
cli_options.configure_logging(args.log_level,
debug_log_file=args.debug_log_file)
seed(args.randomization_seed)
cluster = cli_options.create_cluster(args)
populator = DbPopulator(
db_connection.HIVE if args.use_hive else db_connection.IMPALA)
if command == 'populate':
populator.randomization_seed = args.randomization_seed
populator.cluster = cluster
populator.db_name = args.db_name
populator.min_col_count = args.min_column_count
populator.max_col_count = args.max_column_count
populator.min_row_count = args.min_row_count
populator.max_row_count = args.max_row_count
populator.allowed_storage_formats = args.storage_file_formats.split(
',')
if args.use_hive:
cli_options.add_logging_options(parser)
cli_options.add_cluster_options(parser)
parser.add_argument("-s", "--source-db", required=True,
help="Source DB to load data from.")
parser.add_argument("-t", "--target-db", required=True,
help="Target DB to load data to.")
parser.add_argument("-w", "--workload", choices=['tpch', 'tpcds'],
required=True)
parser.add_argument("--kudu_master", required=True,
help="Address or host name of Kudu master")
# TODO: Automatically set #buckets as a function of cluster nodes and/or
# scale
parser.add_argument("-b", "--buckets", default="9",
help="Number of buckets to partition Kudu tables (only for hash-based).")
parser.add_argument("-v", "--verbose", action='store_true',
help="Print the executed statements.")
parser.add_argument("--clean", action='store_true',
help="Drop all tables in the speficied target database.")
args = parser.parse_args()
cli_options.configure_logging(args.log_level, debug_log_file=args.debug_log_file)
cluster = cli_options.create_cluster(args)
source_db = args.source_db
target_db = args.target_db
buckets = args.buckets
kudu_master = args.kudu_master
workload = args.workload
verbose = args.verbose
if args.clean: clean_data()
load_data()