def __init__(self,
spec_dir,
benchmark_hosts,
result_hosts,
log,
fail_if,
sample_mode):
self.benchmark_hosts = benchmark_hosts
self.sample_mode = sample_mode
self.spec_dir = spec_dir
self.client = clients.client(benchmark_hosts)
self.result_client = clients.client(result_hosts)
self.server_version_info = aio.run(self.client.get_server_version)
self.server_version = parse_version(self.server_version_info['number'])
self.log = log
self.create_result = partial(
Result,
version_info=self.server_version_info
)
if fail_if:
self.fail_if = partial(eval_fail_if, fail_if)
else:
self.fail_if = lambda x: None
if result_hosts:
self.process_result = _result_to_crate(self.log, self.result_client)
else:
self.process_result = log.result
def __init__(self,
spec_dir,
benchmark_hosts,
result_hosts,
log,
fail_if,
sample_mode):
self.benchmark_hosts = benchmark_hosts
self.sample_mode = sample_mode
self.spec_dir = spec_dir
self.client = clients.client(benchmark_hosts)
self.result_client = clients.client(result_hosts)
self.server_version_info = aio.run(self.client.get_server_version)
self.server_version = parse_version(self.server_version_info['number'])
self.log = log
self.create_result = partial(
Result,
version_info=self.server_version_info
)
if fail_if:
self.fail_if = partial(eval_fail_if, fail_if)
else:
self.fail_if = lambda x: None
if result_hosts:
self.process_result = _result_to_crate(self.log, self.result_client)
else:
self.process_result = log.result
def test_reindex(self):
crate_v3 = CrateNode(crate_dir=get_crate('3.x.x'),
keep_data=True,
settings=self.crate_settings)
self._to_stop.append(crate_v3)
crate_v3.start()
with client(crate_v3.http_url) as c:
aio.run(c.execute, "create table t (x int)")
args = (
(1, ),
(2, ),
(3, ),
)
aio.run(c.execute_many, "insert into t (x) values (?)", args)
crate_v3.stop()
self._to_stop.remove(crate_v3)
crate_v4 = CrateNode(crate_dir=get_crate('4.0.3'),
keep_data=True,
settings=self.crate_settings)
self._to_stop.append(crate_v4)
crate_v4.start()
reindex(crate_v4.http_url)
with client(crate_v4.http_url) as c:
result = aio.run(
c.execute,
"SELECT version FROM information_schema.tables WHERE table_name = 't'"
)
version = result['rows'][0][0]
self.assertEqual(version, {'upgraded': None, 'created': '4.0.3'})
cnt = aio.run(c.execute, 'SELECT count(*) FROM t')['rows'][0][0]
self.assertEqual(cnt, 3)
def teardown(*args):
try:
with client(node.http_url) as c:
aio.run(c.execute, 'drop table x.demo')
aio.run(c.execute, 'drop blob table blobtable')
finally:
node.stop()
def insert_from_sql(src_uri=None,
query=None,
fetch_size=100,
concurrency=25,
table=None,
hosts=None,
output_fmt=None):
"""Insert data read from another SQL source into table."""
stats = Stats()
with clients.client(hosts, concurrency=concurrency) as client:
f = partial(aio.measure, stats, client.execute_many)
try:
aio.run(
async_insert_from_sql,
src_uri,
concurrency,
query,
fetch_size,
table,
f
)
except clients.SqlException as e:
raise SystemExit(str(e))
try:
print(format_stats(stats.get(), output_fmt))
except KeyError:
if not stats.sampler.values:
raise SystemExit('No data read from source')
raise
def setup(*args):
with client(node.http_url) as c:
aio.run(
c.execute,
'create table x.demo (id int, name string, country string) \
with (number_of_replicas = 0)'
)
aio.run(c.execute, 'create table y.demo (name text) with (number_of_replicas = 0)')
aio.run(c.execute, 'create blob table blobtable with (number_of_replicas = 0)')
def insert_json(table=None,
bulk_size=1000,
concurrency=25,
hosts=None,
infile=None,
output_fmt=None):
"""Insert JSON lines from a file or stdin into a CrateDB cluster.
If no hosts are specified the statements will be printed.
Args:
table: Target table name.
bulk_size: Bulk size of the insert statements.
concurrency: Number of operations to run concurrently.
hosts: hostname:port pairs of the Crate nodes
"""
if not hosts:
return print_only(infile, table)
queries = (to_insert(table, d) for d in dicts_from_lines(infile))
bulk_queries = as_bulk_queries(queries, bulk_size)
print('Executing inserts: bulk_size={} concurrency={}'.format(
bulk_size, concurrency),
file=sys.stderr)
stats = Stats()
with clients.client(hosts, concurrency=concurrency) as client:
f = partial(aio.measure, stats, client.execute_many)
try:
aio.run_many(f, bulk_queries, concurrency)
except clients.SqlException as e:
raise SystemExit(str(e))
try:
print(format_stats(stats.get(), output_fmt))
except KeyError:
if not stats.sampler.values:
raise SystemExit('No data received via stdin')
raise
def insert_json(table=None,
bulk_size=1000,
concurrency=25,
hosts=None,
output_fmt=None):
"""Insert JSON lines fed into stdin into a Crate cluster.
If no hosts are specified the statements will be printed.
Args:
table: Target table name.
bulk_size: Bulk size of the insert statements.
concurrency: Number of operations to run concurrently.
hosts: hostname:port pairs of the Crate nodes
"""
if not hosts:
return print_only(table)
queries = (to_insert(table, d) for d in dicts_from_stdin())
bulk_queries = as_bulk_queries(queries, bulk_size)
print('Executing inserts: bulk_size={} concurrency={}'.format(
bulk_size, concurrency), file=sys.stderr)
stats = Stats()
with clients.client(hosts, concurrency=concurrency) as client:
f = partial(aio.measure, stats, client.execute_many)
try:
aio.run_many(f, bulk_queries, concurrency)
except clients.SqlException as e:
raise SystemExit(str(e))
try:
print(format_stats(stats.get(), output_fmt))
except KeyError:
if not stats.sampler.values:
raise SystemExit('No data received via stdin')
raise
def reindex(hosts=None):
with clients.client(hosts) as client:
run(_async_reindex, client)
def __init__(self, hosts, concurrency, sample_mode):
self.concurrency = concurrency
self.client = client(hosts, concurrency=concurrency)
self.sampler = get_sampler(sample_mode)
def insert_fake_data(hosts=None,
table=None,
num_records=1e5,
bulk_size=1000,
concurrency=25,
mapping_file=None):
"""Generate random data and insert it into a table.
This will read the table schema and then find suitable random data providers.
Which provider is choosen depends on the column name and data type.
Example:
A column named `name` will map to the `name` provider.
A column named `x` of type int will map to `random_int` because there
is no `x` provider.
Available providers are listed here:
https://faker.readthedocs.io/en/latest/providers.html
Additional providers:
- auto_inc:
Returns unique incrementing numbers.
Automatically used for columns named "id" of type int or long
- geo_point
Returns [<lon>, <lat>]
Automatically used for columns of type geo_point
Args:
hosts: <host>:[<port>] of the Crate node
table: The table name into which the data should be inserted.
Either fully qualified: `<schema>.<table>` or just `<table>`
num_records: Number of records to insert.
Usually a number but expressions like `1e4` work as well.
bulk_size: The bulk size of the insert statements.
concurrency: How many operations to run concurrently.
mapping_file: A JSON file that defines a mapping from column name to
fake-factory provider.
The format is as follows:
{
"column_name": ["provider_with_args", ["arg1", "arg"]],
"x": ["provider_with_args", ["arg1"]],
"y": "provider_without_args"
}
"""
with clients.client(hosts, concurrency=1) as client:
schema, table_name = parse_table(table)
columns = retrieve_columns(client, schema, table_name)
if not columns:
sys.exit('Could not find columns for table "{}"'.format(table))
print('Found schema: ')
print(json.dumps(columns, sort_keys=True, indent=4))
mapping = None
if mapping_file:
mapping = json.load(mapping_file)
bulk_size = min(num_records, bulk_size)
num_inserts = int(math.ceil(num_records / bulk_size))
gen_row = create_row_generator(columns, mapping)
stmt = to_insert('"{schema}"."{table_name}"'.format(**locals()),
columns)[0]
print('Using insert statement: ')
print(stmt)
print('Will make {} requests with a bulk size of {}'.format(
num_inserts, bulk_size))
print('Generating fake data and executing inserts')
q = asyncio.Queue(maxsize=concurrency)
with clients.client(hosts, concurrency=concurrency) as client:
active = [True]
def stop():
asyncio.ensure_future(q.put(None))
active.clear()
loop.remove_signal_handler(signal.SIGINT)
if sys.platform != 'win32':
loop.add_signal_handler(signal.SIGINT, stop)
bulk_seq = _bulk_size_generator(num_records, bulk_size, active)
with ThreadPoolExecutor() as e:
tasks = asyncio.gather(
_gen_data_and_insert(q, e, client, stmt, gen_row, bulk_seq),
consume(q, total=num_inserts))
loop.run_until_complete(tasks)
def __init__(self, hosts, concurrency, sample_mode):
self.concurrency = concurrency
self.client = client(hosts, concurrency=concurrency)
self.sampler = get_sampler(sample_mode)
def insert_fake_data(hosts=None,
table=None,
num_records=1e5,
bulk_size=1000,
concurrency=25,
mapping_file=None):
"""Generate random data and insert it into a table.
This will read the table schema and then find suitable random data providers.
Which provider is choosen depends on the column name and data type.
Example:
A column named `name` will map to the `name` provider.
A column named `x` of type int will map to `random_int` because there
is no `x` provider.
Available providers are listed here:
https://faker.readthedocs.io/en/latest/providers.html
Additional providers:
- auto_inc:
Returns unique incrementing numbers.
Automatically used for columns named "id" of type int or long
- geo_point
Returns [<lon>, <lat>]
Automatically used for columns of type geo_point
Args:
hosts: <host>:[<port>] of the Crate node
table: The table name into which the data should be inserted.
Either fully qualified: `<schema>.<table>` or just `<table>`
num_records: Number of records to insert.
Usually a number but expressions like `1e4` work as well.
bulk_size: The bulk size of the insert statements.
concurrency: How many operations to run concurrently.
mapping_file: A JSON file that defines a mapping from column name to
fake-factory provider.
The format is as follows:
{
"column_name": ["provider_with_args", ["arg1", "arg"]],
"x": ["provider_with_args", ["arg1"]],
"y": "provider_without_args"
}
"""
with clients.client(hosts, concurrency=1) as client:
schema, table_name = parse_table(table)
columns = retrieve_columns(client, schema, table_name)
if not columns:
sys.exit('Could not find columns for table "{}"'.format(table))
print('Found schema: ')
print(json.dumps(columns, sort_keys=True, indent=4))
mapping = None
if mapping_file:
mapping = json.load(mapping_file)
bulk_size = min(num_records, bulk_size)
num_inserts = int(math.ceil(num_records / bulk_size))
gen_row = create_row_generator(columns, mapping)
stmt = to_insert('"{schema}"."{table_name}"'.format(**locals()), columns)[0]
print('Using insert statement: ')
print(stmt)
print('Will make {} requests with a bulk size of {}'.format(
num_inserts, bulk_size))
print('Generating fake data and executing inserts')
q = asyncio.Queue(maxsize=concurrency)
with clients.client(hosts, concurrency=concurrency) as client:
active = [True]
def stop():
asyncio.ensure_future(q.put(None))
active.clear()
loop.remove_signal_handler(signal.SIGINT)
if sys.platform != 'win32':
loop.add_signal_handler(signal.SIGINT, stop)
bulk_seq = _bulk_size_generator(num_records, bulk_size, active)
with ThreadPoolExecutor() as e:
tasks = asyncio.gather(
_gen_data_and_insert(q, e, client, stmt, gen_row, bulk_seq),
consume(q, total=num_inserts)
)
loop.run_until_complete(tasks)