def _populate_secondary_fs_partitions(self, db_name):
# This directory may already exist. So we needn't mind if this call fails.
call(["hadoop", "fs", "-mkdir", get_secondary_fs_path("/multi_fs_tests/")], shell=False)
check_call(["hadoop", "fs", "-mkdir",
get_secondary_fs_path("/multi_fs_tests/%s.db/" % db_name)], shell=False)
check_call(["hadoop", "fs", "-cp", "/test-warehouse/alltypes_parquet/",
get_secondary_fs_path("/multi_fs_tests/%s.db/" % db_name)], shell=False)
check_call(["hadoop", "fs", "-cp", "/test-warehouse/tinytable/",
get_secondary_fs_path("/multi_fs_tests/%s.db/" % db_name)], shell=False)
def test_multiple_filesystems(self, vector, unique_database):
try:
self._populate_secondary_fs_partitions(unique_database)
self.run_test_case('QueryTest/multiple-filesystems', vector, use_db=unique_database)
finally:
# We delete this from the secondary filesystem here because the database was created
# in HDFS but the queries will create this path in the secondary FS as well. So
# dropping the database will not delete the directory in the secondary FS.
check_call(["hadoop", "fs", "-rm", "-r",
get_secondary_fs_path("/multi_fs_tests/%s.db/" % unique_database)], shell=False)
def _populate_secondary_fs_partitions(self, db_name):
# This directory may already exist. So we needn't mind if this call fails.
call([
"hadoop", "fs", "-mkdir",
get_secondary_fs_path("/multi_fs_tests/")
],
shell=False)
check_call([
"hadoop", "fs", "-mkdir",
get_secondary_fs_path("/multi_fs_tests/%s.db/" % db_name)
],
shell=False)
self.filesystem_client.copy("/test-warehouse/alltypes_parquet/",
get_secondary_fs_path(
"/multi_fs_tests/%s.db/" % db_name),
overwrite=True)
self.filesystem_client.copy("/test-warehouse/tinytable/",
get_secondary_fs_path(
"/multi_fs_tests/%s.db/" % db_name),
overwrite=True)
def test_multiple_filesystems(self, vector, unique_database):
try:
self._populate_secondary_fs_partitions(unique_database)
self.run_test_case('QueryTest/multiple-filesystems',
vector,
use_db=unique_database)
finally:
# We delete this from the secondary filesystem here because the database was created
# in HDFS but the queries will create this path in the secondary FS as well. So
# dropping the database will not delete the directory in the secondary FS.
check_call([
"hadoop", "fs", "-rm", "-r",
get_secondary_fs_path(
"/multi_fs_tests/%s.db/" % unique_database)
],
shell=False)