def test_bucket_delete_with_objects(self, mcg_obj, interface, awscli_pod):
"""
Negative test with deletion of bucket has objects stored in.
"""
bucket_map = {'s3': S3Bucket, 'oc': OCBucket, 'cli': CLIBucket}
bucketname = create_unique_resource_name(
resource_description='bucket', resource_type=interface.lower())
try:
bucket = bucket_map[interface.lower()](mcg_obj, bucketname)
logger.info(f"aws s3 endpoint is {mcg_obj.s3_endpoint}")
logger.info(f"aws region is {mcg_obj.region}")
data_dir = '/data'
full_object_path = f"s3://{bucketname}"
helpers.retrieve_test_objects_to_pod(awscli_pod, data_dir)
helpers.sync_object_directory(awscli_pod, data_dir,
full_object_path, mcg_obj)
logger.info(f"Deleting bucket: {bucketname}")
if interface == "S3":
try:
s3_del = mcg_obj.s3_resource.Bucket(bucketname).delete()
assert not s3_del, (
"Unexpected s3 delete non-empty OBC succeed")
except botocore.exceptions.ClientError as err:
assert "BucketNotEmpty" in str(err), (
"Couldn't verify delete non-empty OBC with s3")
logger.info(
f"Delete non-empty OBC {bucketname} failed as expected"
)
finally:
bucket.delete()
def test_check_object_integrity(self, mcg_obj, awscli_pod, bucket_factory):
"""
Test object integrity using md5sum
"""
bucketname = bucket_factory(1)[0].name
original_dir = "/original"
result_dir = "/result"
awscli_pod.exec_cmd_on_pod(command=f'mkdir {result_dir}')
# Retrieve a list of all objects on the test-objects bucket and
# downloads them to the pod
full_object_path = f"s3://{bucketname}"
downloaded_files = helpers.retrieve_test_objects_to_pod(
awscli_pod, original_dir)
# Write all downloaded objects to the new bucket
helpers.sync_object_directory(awscli_pod, original_dir,
full_object_path, mcg_obj)
# Retrieve all objects from MCG bucket to result dir in Pod
logger.info(f'Downloading all objects from MCG bucket to awscli pod')
helpers.sync_object_directory(awscli_pod, full_object_path, result_dir,
mcg_obj)
# Checksum is compared between original and result object
for obj in downloaded_files:
assert mcg_obj.verify_s3_object_integrity(
original_object_path=f'{original_dir}/{obj}',
result_object_path=f'{result_dir}/{obj}',
awscli_pod=awscli_pod
), 'Checksum comparision between original and result object failed'
def test_empty_file_integrity(self, mcg_obj, awscli_pod, bucket_factory):
"""
Test write empty files to bucket and check integrity
"""
original_dir = '/data'
result_dir = "/result"
bucketname = bucket_factory(1)[0].name
full_object_path = f"s3://{bucketname}"
# Touch create 1000 empty files in pod
awscli_pod.exec_cmd_on_pod(
command=f'mkdir {original_dir} {result_dir}')
command = "for i in $(seq 1 100); do touch /data/test$i; done"
awscli_pod.exec_sh_cmd_on_pod(command=command, sh='sh')
# Write all empty objects to the new bucket
helpers.sync_object_directory(awscli_pod, original_dir,
full_object_path, mcg_obj)
# Retrieve all objects from MCG bucket to result dir in Pod
logger.info(f'Downloading objects from MCG bucket to awscli pod')
helpers.sync_object_directory(awscli_pod, full_object_path, result_dir,
mcg_obj)
# Checksum is compared between original and result object
for obj in ('test' + str(i + 1) for i in range(100)):
assert mcg_obj.verify_s3_object_integrity(
original_object_path=f'{original_dir}/{obj}',
result_object_path=f'{result_dir}/{obj}',
awscli_pod=awscli_pod
), 'Checksum comparision between original and result object failed'
def test_fill_bucket(
mcg_obj_session,
awscli_pod_session,
multiregion_mirror_setup_session
):
"""
Test multi-region bucket creation using the S3 SDK. Fill the bucket for
upgrade testing.
"""
(
bucket,
backingstore1,
backingstore2
) = multiregion_mirror_setup_session
mcg_bucket_path = f's3://{bucket.name}'
# Download test objects from the public bucket
awscli_pod_session.exec_cmd_on_pod(
command=f'mkdir {LOCAL_TESTOBJS_DIR_PATH}'
)
DOWNLOADED_OBJS = retrieve_test_objects_to_pod(
awscli_pod_session, LOCAL_TESTOBJS_DIR_PATH
)
logger.info('Uploading all pod objects to MCG bucket')
# Upload test objects to the NooBucket 3 times
for i in range(3):
sync_object_directory(
awscli_pod_session,
LOCAL_TESTOBJS_DIR_PATH,
f'{mcg_bucket_path}/{i}/',
mcg_obj_session
)
mcg_obj_session.check_if_mirroring_is_done(bucket.name)
assert bucket.status == constants.STATUS_BOUND
# Retrieve all objects from MCG bucket to result dir in Pod
sync_object_directory(
awscli_pod_session,
mcg_bucket_path,
LOCAL_TEMP_PATH,
mcg_obj_session
)
assert bucket.status == constants.STATUS_BOUND
# Checksum is compared between original and result object
for obj in DOWNLOADED_OBJS:
for i in range(3):
assert mcg_obj_session.verify_s3_object_integrity(
original_object_path=f'{LOCAL_TESTOBJS_DIR_PATH}/{obj}',
result_object_path=f'{LOCAL_TEMP_PATH}/{i}/{obj}',
awscli_pod=awscli_pod_session
), 'Checksum comparison between original and result object failed'
assert bucket.status == constants.STATUS_BOUND
def test_check_multi_object_integrity(self, mcg_obj, awscli_pod,
bucket_factory, amount, file_type):
"""
Test write multiple files to bucket and check integrity
"""
original_dir = "/original"
result_dir = "/result"
if file_type == 'large':
public_bucket = PUBLIC_BUCKET
obj_key = LARGE_FILE_KEY
elif file_type == 'small':
public_bucket = constants.TEST_FILES_BUCKET
obj_key = 'random1.txt'
elif file_type == 'large_small':
public_bucket = PUBLIC_BUCKET
obj_key = LARGE_FILE_KEY.rsplit('/', 1)[0]
# Download the file to pod
awscli_pod.exec_cmd_on_pod(
command=f'mkdir {original_dir} {result_dir}')
public_s3 = boto3.client('s3')
download_files = []
# Use obj_key as prefix to download multiple files for large_small
# case, it also works with single file
for obj in public_s3.list_objects(Bucket=public_bucket,
Prefix=obj_key).get('Contents'):
# Skip the extra file in large file type
if file_type == 'large' and obj["Key"] != obj_key:
continue
logger.info(
f'Downloading {obj["Key"]} from AWS bucket {public_bucket}')
command = f'wget -P {original_dir} '
command += f'https://{public_bucket}.s3.amazonaws.com/{obj["Key"]}'
awscli_pod.exec_cmd_on_pod(command=command)
download_files.append(obj['Key'].split('/')[-1])
# Write downloaded objects to the new bucket and check integrity
bucketname = bucket_factory(1)[0].name
base_path = f"s3://{bucketname}"
for i in range(amount):
full_object_path = base_path + f"/{i}/"
helpers.sync_object_directory(awscli_pod, original_dir,
full_object_path, mcg_obj)
# Retrieve all objects from MCG bucket to result dir in Pod
logger.info(f'Downloading objects from MCG bucket to awscli pod')
helpers.sync_object_directory(awscli_pod, full_object_path,
result_dir, mcg_obj)
# Checksum is compared between original and result object
for obj in download_files:
assert mcg_obj.verify_s3_object_integrity(
original_object_path=f'{original_dir}/{obj}',
result_object_path=f'{result_dir}/{obj}',
awscli_pod=awscli_pod
), ('Checksum comparision between original and result object '
'failed')
def test_write_multi_files_to_bucket(self, mcg_obj, awscli_pod,
bucket_factory, amount, file_type):
"""
Test write multiple files to bucket
"""
data_dir = '/data'
if file_type == 'large':
public_bucket = PUBLIC_BUCKET
obj_key = LARGE_FILE_KEY
elif file_type == 'small':
public_bucket = constants.TEST_FILES_BUCKET
obj_key = 'random1.txt'
elif file_type == 'large_small':
public_bucket = PUBLIC_BUCKET
obj_key = LARGE_FILE_KEY.rsplit('/', 1)[0]
# Download the file to pod
awscli_pod.exec_cmd_on_pod(command=f'mkdir {data_dir}')
public_s3_client = retrieve_anon_s3_resource().meta.client
download_files = []
# Use obj_key as prefix to download multiple files for large_small
# case, it also works with single file
for obj in public_s3_client.list_objects(
Bucket=public_bucket, Prefix=obj_key).get('Contents'):
# Skip the extra file in large file type
if file_type == 'large' and obj["Key"] != obj_key:
continue
logger.info(
f'Downloading {obj["Key"]} from AWS bucket {public_bucket}')
download_obj_cmd = f'cp s3://{public_bucket}/{obj["Key"]} {data_dir}'
awscli_pod.exec_cmd_on_pod(
command=craft_s3_command(download_obj_cmd),
out_yaml_format=False)
download_files.append(obj['Key'])
# Write all downloaded objects to the new bucket
bucketname = bucket_factory(1)[0].name
base_path = f"s3://{bucketname}"
for i in range(amount):
full_object_path = base_path + f"/{i}/"
helpers.sync_object_directory(awscli_pod, data_dir,
full_object_path, mcg_obj)
obj_list = list(
obj.key.split('/')[-1]
for obj in mcg_obj.s3_list_all_objects_in_bucket(bucketname))
# Check total copy files amount match
if file_type == 'large_small':
assert len(obj_list) == 2 * amount, (
"Total file amount does not match")
else:
assert len(obj_list) == amount, "Total file amount does not match"
# Check deduplicate set is same
test_set = set([i.split('/')[-1] for i in download_files])
assert test_set == set(obj_list), "File name set does not match"
def test_noobaa_postupgrade(
mcg_obj_session,
awscli_pod_session,
multiregion_mirror_setup_session
):
"""
Check bucket data and remove resources created in 'test_fill_bucket'.
"""
(
bucket,
backingstore1,
backingstore2
) = multiregion_mirror_setup_session
mcg_bucket_path = f's3://{bucket.name}'
# Checksum is compared between original and result object
for obj in DOWNLOADED_OBJS:
assert mcg_obj_session.verify_s3_object_integrity(
original_object_path=f'{LOCAL_TESTOBJS_DIR_PATH}/{obj}',
result_object_path=f'{LOCAL_TEMP_PATH}/{obj}',
awscli_pod=awscli_pod_session
), 'Checksum comparision between original and result object failed'
assert bucket.status == constants.STATUS_BOUND
# Clean up the temp dir
awscli_pod_session.exec_cmd_on_pod(
command=f'sh -c \"rm -rf {LOCAL_TEMP_PATH}/*\"'
)
mcg_obj_session.check_backingstore_state(
'backing-store-' + backingstore1['name'],
BS_OPTIMAL,
timeout=360
)
mcg_obj_session.check_backingstore_state(
'backing-store-' + backingstore2['name'],
BS_OPTIMAL,
timeout=360
)
assert bucket.status == constants.STATUS_BOUND
# Verify integrity of A
# Retrieve all objects from MCG bucket to result dir in Pod
sync_object_directory(
awscli_pod_session,
mcg_bucket_path,
LOCAL_TEMP_PATH,
mcg_obj_session
)
assert bucket.status == constants.STATUS_BOUND
def test_data_reduction(self, mcg_obj, awscli_pod, bucket_factory):
"""
Test data reduction mechanics
"""
# TODO: Privatize test bucket
download_dir = '/aws/downloaded'
helpers.retrieve_test_objects_to_pod(awscli_pod, download_dir)
bucket = bucket_factory(1)[0]
bucketname = bucket.name
full_object_path = f"s3://{bucketname}"
helpers.sync_object_directory(awscli_pod, download_dir,
full_object_path, mcg_obj)
assert mcg_obj.check_data_reduction(bucketname), (
'Data reduction did not work as anticipated.')
def test_multipart_upload_operations(self, mcg_obj, awscli_pod,
bucket_factory):
"""
Test Multipart upload operations on bucket and verifies the integrity of the downloaded object
"""
bucket, key, origin_dir, res_dir, object_path, parts = setup(
awscli_pod, bucket_factory)
# Abort all Multipart Uploads for this Bucket (optional, for starting over)
logger.info(f'Aborting any Multipart Upload on bucket:{bucket}')
mcg_obj.abort_multipart_upload(bucket, key)
# Create & list Multipart Upload on the Bucket
logger.info(
f'Initiating Multipart Upload on Bucket: {bucket} with Key {key}')
upload_id = mcg_obj.create_multipart_upload(bucket, key)
logger.info(
f'Listing the Multipart Upload : {mcg_obj.list_multipart_upload(bucket)}'
)
# Uploading individual parts to the Bucket
logger.info(f'Uploading individual parts to the bucket {bucket}')
uploaded_parts = helpers.upload_parts(mcg_obj, awscli_pod, bucket, key,
res_dir, upload_id, parts)
# Listing the Uploaded parts
logger.info(
f'Listing the individual parts : {mcg_obj.list_uploaded_parts(bucket, key, upload_id)}'
)
# Completing the Multipart Upload
logger.info(f'Completing the Multipart Upload on bucket: {bucket}')
logger.info(
mcg_obj.complete_multipart_upload(bucket, key, upload_id,
uploaded_parts))
# Checksum Validation: Downloading the object after completing Multipart Upload and verifying its integrity
logger.info(
f'Downloading the completed multipart object from MCG bucket to awscli pod'
)
helpers.sync_object_directory(awscli_pod, object_path, res_dir,
mcg_obj)
assert mcg_obj.verify_s3_object_integrity(
original_object_path=f'{origin_dir}/{key}',
result_object_path=f'{res_dir}/{key}',
awscli_pod=awscli_pod
), 'Checksum comparision between original and result object failed'
def test_write_file_to_bucket(self, mcg_obj, awscli_pod, bucket_factory):
"""
Test object IO using the S3 SDK
"""
# Retrieve a list of all objects on the test-objects bucket and
# downloads them to the pod
data_dir = '/data'
bucketname = bucket_factory(1)[0].name
full_object_path = f"s3://{bucketname}"
downloaded_files = helpers.retrieve_test_objects_to_pod(
awscli_pod, data_dir)
# Write all downloaded objects to the new bucket
helpers.sync_object_directory(awscli_pod, data_dir, full_object_path,
mcg_obj)
assert set(downloaded_files).issubset(
obj.key
for obj in mcg_obj.s3_list_all_objects_in_bucket(bucketname))
def test_write_empty_file_to_bucket(self, mcg_obj, awscli_pod,
bucket_factory):
"""
Test write empty files to bucket
"""
data_dir = '/data'
bucketname = bucket_factory(1)[0].name
full_object_path = f"s3://{bucketname}"
# Touch create 1000 empty files in pod
awscli_pod.exec_cmd_on_pod(command=f'mkdir {data_dir}')
command = "for i in $(seq 1 1000); do touch /data/test$i; done"
awscli_pod.exec_sh_cmd_on_pod(command=command, sh='sh')
# Write all empty objects to the new bucket
helpers.sync_object_directory(awscli_pod, data_dir, full_object_path,
mcg_obj)
obj_set = set(
obj.key
for obj in mcg_obj.s3_list_all_objects_in_bucket(bucketname))
test_set = set('test' + str(i + 1) for i in range(1000))
assert test_set == obj_set, "File name set does not match"
def test_fill_bucket(mcg_obj_session, awscli_pod_session,
multiregion_mirror_setup_session):
"""
Test multi-region bucket creation using the S3 SDK. Fill the bucket for
upgrade testing.
"""
(bucket, backingstore1, backingstore2) = multiregion_mirror_setup_session
mcg_bucket_path = f's3://{bucket.name}'
# Download test objects from the public bucket
awscli_pod_session.exec_cmd_on_pod(
command=f'mkdir {LOCAL_TESTOBJS_DIR_PATH}')
test_objects = boto3.resource('s3').Bucket(
constants.TEST_FILES_BUCKET).objects.all()
for obj in test_objects:
logger.info(f'Downloading {obj.key} from AWS test bucket')
awscli_pod_session.exec_cmd_on_pod(command=(
f'sh -c "wget -P {LOCAL_TESTOBJS_DIR_PATH} '
f'https://{constants.TEST_FILES_BUCKET}.s3.amazonaws.com/{obj.key}"'
))
DOWNLOADED_OBJS.append(f'{obj.key}')
# Use 3x time more objects than there is in test objects pod
for i in range(2):
awscli_pod_session.exec_cmd_on_pod(
command=f'sh -c "'
f'cp {LOCAL_TESTOBJS_DIR_PATH}/{obj.key} '
f'{LOCAL_TESTOBJS_DIR_PATH}/{obj.key}.{i}"')
DOWNLOADED_OBJS.append(f'{obj.key}.{i}')
logger.info(f'Uploading all pod objects to MCG bucket')
sync_object_directory(awscli_pod_session,
's3://' + constants.TEST_FILES_BUCKET,
LOCAL_TESTOBJS_DIR_PATH)
# Upload test objects to the NooBucket
sync_object_directory(awscli_pod_session, LOCAL_TESTOBJS_DIR_PATH,
mcg_bucket_path, mcg_obj_session)
mcg_obj_session.check_if_mirroring_is_done(bucket.name)
assert bucket.status == constants.STATUS_BOUND
# Retrieve all objects from MCG bucket to result dir in Pod
sync_object_directory(awscli_pod_session, mcg_bucket_path, LOCAL_TEMP_PATH,
mcg_obj_session)
assert bucket.status == constants.STATUS_BOUND
# Checksum is compared between original and result object
for obj in DOWNLOADED_OBJS:
assert mcg_obj_session.verify_s3_object_integrity(
original_object_path=f'{LOCAL_TESTOBJS_DIR_PATH}/{obj}',
result_object_path=f'{LOCAL_TEMP_PATH}/{obj}',
awscli_pod=awscli_pod_session
), 'Checksum comparision between original and result object failed'
assert bucket.status == constants.STATUS_BOUND
def test_s3_bucket_delete_1t_objects(self, mcg_obj, awscli_pod):
"""
Test with deletion of bucket has 1T objects stored in.
"""
bucketname = create_unique_resource_name(
resource_description='bucket', resource_type='s3'
)
try:
bucket = S3Bucket(mcg_obj, bucketname)
logger.info(f"aws s3 endpoint is {mcg_obj.s3_endpoint}")
logger.info(f"aws region is {mcg_obj.region}")
data_dir = '/data'
helpers.retrieve_test_objects_to_pod(awscli_pod, data_dir)
# Sync downloaded objects dir to the new bucket, sync to 3175
# virtual dirs. With each dir around 315MB, and 3175 dirs will
# reach targed 1TB data.
logger.info('Writing objects to bucket')
for i in range(3175):
full_object_path = f"s3://{bucketname}/{i}/"
helpers.sync_object_directory(
awscli_pod, data_dir, full_object_path, mcg_obj
)
# Delete bucket content use aws rm with --recursive option.
# The object_versions.delete function does not work with objects
# exceeds 1000.
start = timeit.default_timer()
helpers.rm_object_recursive(awscli_pod, bucketname, mcg_obj)
bucket.delete()
stop = timeit.default_timer()
gap = (stop - start) // 60 % 60
if gap > 10:
assert False, "Failed to delete s3 bucket within 10 minutes"
finally:
if mcg_obj.s3_verify_bucket_exists(bucketname):
helpers.rm_object_recursive(awscli_pod, bucketname, mcg_obj)
mcg_obj.s3_resource.Bucket(bucketname).delete()
def test_write_to_bucket_rbd_cephfs(self, verify_rgw_restart_count,
setup_rbd_cephfs_pods, mcg_obj,
awscli_pod, bucket_factory):
"""
Test RGW restarts after running s3, rbd and cephfs IOs in parallel
"""
bucketname = bucket_factory(1)[0].name
full_object_path = f"s3://{bucketname}"
target_dir = '/data/'
helpers.retrieve_test_objects_to_pod(awscli_pod, target_dir)
with ThreadPoolExecutor() as p:
p.submit(pod_io, setup_rbd_cephfs_pods)
p.submit(
helpers.sync_object_directory(awscli_pod, target_dir,
full_object_path, mcg_obj))
def test_multiregion_mirror(self, mcg_obj, awscli_pod,
multiregion_mirror_setup):
"""
Test multi-region bucket creation using the S3 SDK
"""
bucket, backingstore1, backingstore2 = multiregion_mirror_setup
bucket_name = bucket.name
# Download test objects from the public bucket
downloaded_objs = retrieve_test_objects_to_pod(awscli_pod,
'/aws/original/')
logger.info(f'Uploading all pod objects to MCG bucket')
local_testobjs_dir_path = '/aws/original'
local_temp_path = '/aws/temp'
mcg_bucket_path = f's3://{bucket_name}'
sync_object_directory(awscli_pod,
's3://' + constants.TEST_FILES_BUCKET,
local_testobjs_dir_path)
# Upload test objects to the NooBucket
sync_object_directory(awscli_pod, local_testobjs_dir_path,
mcg_bucket_path, mcg_obj)
mcg_obj.check_if_mirroring_is_done(bucket_name)
# Bring bucket A down
mcg_obj.toggle_aws_bucket_readwrite(backingstore1['name'])
mcg_obj.check_backingstore_state(
'backing-store-' + backingstore1['name'], BS_AUTH_FAILED)
# Verify integrity of B
# Retrieve all objects from MCG bucket to result dir in Pod
sync_object_directory(awscli_pod, mcg_bucket_path, local_temp_path,
mcg_obj)
# Checksum is compared between original and result object
for obj in downloaded_objs:
assert mcg_obj.verify_s3_object_integrity(
original_object_path=f'{local_testobjs_dir_path}/{obj}',
result_object_path=f'{local_temp_path}/{obj}',
awscli_pod=awscli_pod
), 'Checksum comparision between original and result object failed'
# Clean up the temp dir
awscli_pod.exec_cmd_on_pod(
command=f'sh -c \"rm -rf {local_temp_path}/*\"')
# Bring B down, bring A up
logger.info('Blocking bucket B')
mcg_obj.toggle_aws_bucket_readwrite(backingstore2['name'])
logger.info('Freeing bucket A')
mcg_obj.toggle_aws_bucket_readwrite(backingstore1['name'], block=False)
mcg_obj.check_backingstore_state(
'backing-store-' + backingstore1['name'], BS_OPTIMAL)
mcg_obj.check_backingstore_state(
'backing-store-' + backingstore2['name'], BS_AUTH_FAILED)
# Verify integrity of A
# Retrieve all objects from MCG bucket to result dir in Pod
sync_object_directory(awscli_pod, mcg_bucket_path, local_temp_path,
mcg_obj)
# Checksum is compared between original and result object
for obj in downloaded_objs:
assert mcg_obj.verify_s3_object_integrity(
original_object_path=f'{local_testobjs_dir_path}/{obj}',
result_object_path=f'{local_temp_path}/{obj}',
awscli_pod=awscli_pod
), 'Checksum comparision between original and result object failed'
# Bring B up
mcg_obj.toggle_aws_bucket_readwrite(backingstore2['name'], block=False)
mcg_obj.check_backingstore_state(
'backing-store-' + backingstore2['name'], BS_OPTIMAL)
