e1140666d6
This patch adds methods to increase the partition power of an existing object ring without downtime for the users using a 3-step process. Data won't be moved to other nodes; objects using the new increased partition power will be located on the same device and are hardlinked to avoid data movement. 1. A new setting "next_part_power" will be added to the rings, and once the proxy server reloaded the rings it will send this value to the object servers on any write operation. Object servers will now create a hard-link in the new location to the original DiskFile object. Already existing data will be relinked using a new tool in the new locations using hardlinks. 2. The actual partition power itself will be increased. Servers will now use the new partition power to read from and write to. No longer required hard links in the old object location have to be removed now by the relinker tool; the relinker tool reads the next_part_power setting to find object locations that need to be cleaned up. 3. The "next_part_power" flag will be removed. This mostly implements the spec in [1]; however it's not using an "epoch" as described there. The idea of the epoch was to store data using different partition powers in their own namespace to avoid conflicts with auditors and replicators as well as being able to abort such an operation and just remove the new tree. This would require some heavy change of the on-disk data layout, and other object-server implementations would be required to adopt this scheme too. Instead the object-replicator is now aware that there is a partition power increase in progress and will skip replication of data in that storage policy; the relinker tool should be simply run and afterwards the partition power will be increased. This shouldn't take that much time (it's only walking the filesystem and hardlinking); impact should be low therefore. The relinker should be run on all storage nodes at the same time in parallel to decrease the required time (though this is not mandatory). Failures during relinking should not affect cluster operations - relinking can be even aborted manually and restarted later. Auditors are not quarantining objects written to a path with a different partition power and therefore working as before (though they are reading each object twice in the worst case before the no longer needed hard links are removed). Co-Authored-By: Alistair Coles <alistair.coles@hpe.com> Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> [1] https://specs.openstack.org/openstack/swift-specs/specs/in_progress/ increasing_partition_power.html Change-Id: I7d6371a04f5c1c4adbb8733a71f3c177ee5448bb
200 lines
7.3 KiB
Python
Executable File
200 lines
7.3 KiB
Python
Executable File
#!/usr/bin/env python
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
|
# implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
import os
|
|
|
|
from errno import EEXIST
|
|
from shutil import copyfile
|
|
from tempfile import mkstemp
|
|
from time import time
|
|
from unittest import main
|
|
from uuid import uuid4
|
|
|
|
from swiftclient import client
|
|
|
|
from swift.cli.relinker import relink, cleanup
|
|
from swift.common.manager import Manager
|
|
from swift.common.ring import RingBuilder
|
|
from swift.common.utils import replace_partition_in_path
|
|
from swift.obj.diskfile import get_data_dir
|
|
from test.probe.common import ECProbeTest, ProbeTest, ReplProbeTest
|
|
|
|
|
|
class TestPartPowerIncrease(ProbeTest):
|
|
def setUp(self):
|
|
super(TestPartPowerIncrease, self).setUp()
|
|
_, self.ring_file_backup = mkstemp()
|
|
_, self.builder_file_backup = mkstemp()
|
|
self.ring_file = self.object_ring.serialized_path
|
|
self.builder_file = self.ring_file.replace('ring.gz', 'builder')
|
|
copyfile(self.ring_file, self.ring_file_backup)
|
|
copyfile(self.builder_file, self.builder_file_backup)
|
|
# In case the test user is not allowed to write rings
|
|
self.assertTrue(os.access('/etc/swift', os.W_OK))
|
|
self.assertTrue(os.access('/etc/swift/backups', os.W_OK))
|
|
self.assertTrue(os.access('/etc/swift/object.builder', os.W_OK))
|
|
self.assertTrue(os.access('/etc/swift/object.ring.gz', os.W_OK))
|
|
# Ensure the test object will be erasure coded
|
|
self.data = ' ' * getattr(self.policy, 'ec_segment_size', 1)
|
|
|
|
self.devices = [
|
|
self.device_dir('object', {'ip': ip, 'port': port, 'device': ''})
|
|
for ip, port in set((dev['ip'], dev['port'])
|
|
for dev in self.object_ring.devs)]
|
|
|
|
def tearDown(self):
|
|
# Keep a backup copy of the modified .builder file
|
|
backup_dir = os.path.join(
|
|
os.path.dirname(self.builder_file), 'backups')
|
|
try:
|
|
os.mkdir(backup_dir)
|
|
except OSError as err:
|
|
if err.errno != EEXIST:
|
|
raise
|
|
backup_name = (os.path.join(
|
|
backup_dir,
|
|
'%d.probe.' % time() + os.path.basename(self.builder_file)))
|
|
copyfile(self.builder_file, backup_name)
|
|
|
|
# Restore original ring
|
|
os.system('sudo mv %s %s' % (
|
|
self.ring_file_backup, self.ring_file))
|
|
os.system('sudo mv %s %s' % (
|
|
self.builder_file_backup, self.builder_file))
|
|
|
|
def _find_objs_ondisk(self, container, obj):
|
|
locations = []
|
|
opart, onodes = self.object_ring.get_nodes(
|
|
self.account, container, obj)
|
|
for node in onodes:
|
|
start_dir = os.path.join(
|
|
self.device_dir('object', node),
|
|
get_data_dir(self.policy),
|
|
str(opart))
|
|
for root, dirs, files in os.walk(start_dir):
|
|
for filename in files:
|
|
if filename.endswith('.data'):
|
|
locations.append(os.path.join(root, filename))
|
|
return locations
|
|
|
|
def _test_main(self, cancel=False):
|
|
container = 'container-%s' % uuid4()
|
|
obj = 'object-%s' % uuid4()
|
|
obj2 = 'object-%s' % uuid4()
|
|
|
|
# Create container
|
|
headers = {'X-Storage-Policy': self.policy.name}
|
|
client.put_container(self.url, self.token, container, headers=headers)
|
|
|
|
# Create a new object
|
|
client.put_object(self.url, self.token, container, obj, self.data)
|
|
client.head_object(self.url, self.token, container, obj)
|
|
|
|
# Prepare partition power increase
|
|
builder = RingBuilder.load(self.builder_file)
|
|
builder.prepare_increase_partition_power()
|
|
builder.save(self.builder_file)
|
|
ring_data = builder.get_ring()
|
|
ring_data.save(self.ring_file)
|
|
|
|
# Ensure the proxy uses the changed ring
|
|
Manager(['proxy']).restart()
|
|
|
|
# Ensure object is still accessible
|
|
client.head_object(self.url, self.token, container, obj)
|
|
|
|
# Relink existing objects
|
|
for device in self.devices:
|
|
self.assertEqual(0, relink(skip_mount_check=True, devices=device))
|
|
|
|
# Create second object after relinking and ensure it is accessible
|
|
client.put_object(self.url, self.token, container, obj2, self.data)
|
|
client.head_object(self.url, self.token, container, obj2)
|
|
|
|
# Remember the original object locations
|
|
org_locations = self._find_objs_ondisk(container, obj)
|
|
org_locations += self._find_objs_ondisk(container, obj2)
|
|
|
|
# Remember the new object locations
|
|
new_locations = []
|
|
for loc in org_locations:
|
|
new_locations.append(replace_partition_in_path(
|
|
str(loc), self.object_ring.part_power + 1))
|
|
|
|
# Overwrite existing object - to ensure that older timestamp files
|
|
# will be cleaned up properly later
|
|
client.put_object(self.url, self.token, container, obj, self.data)
|
|
|
|
# Ensure objects are still accessible
|
|
client.head_object(self.url, self.token, container, obj)
|
|
client.head_object(self.url, self.token, container, obj2)
|
|
|
|
# Increase partition power
|
|
builder = RingBuilder.load(self.builder_file)
|
|
if not cancel:
|
|
builder.increase_partition_power()
|
|
else:
|
|
builder.cancel_increase_partition_power()
|
|
builder.save(self.builder_file)
|
|
ring_data = builder.get_ring()
|
|
ring_data.save(self.ring_file)
|
|
|
|
# Ensure the proxy uses the changed ring
|
|
Manager(['proxy']).restart()
|
|
|
|
# Ensure objects are still accessible
|
|
client.head_object(self.url, self.token, container, obj)
|
|
client.head_object(self.url, self.token, container, obj2)
|
|
|
|
# Overwrite existing object - to ensure that older timestamp files
|
|
# will be cleaned up properly later
|
|
client.put_object(self.url, self.token, container, obj, self.data)
|
|
|
|
# Cleanup old objects in the wrong location
|
|
for device in self.devices:
|
|
self.assertEqual(0, cleanup(skip_mount_check=True, devices=device))
|
|
|
|
# Ensure objects are still accessible
|
|
client.head_object(self.url, self.token, container, obj)
|
|
client.head_object(self.url, self.token, container, obj2)
|
|
|
|
# Ensure data in old or relinked object locations is removed
|
|
if not cancel:
|
|
for fn in org_locations:
|
|
self.assertFalse(os.path.exists(fn))
|
|
else:
|
|
for fn in new_locations:
|
|
self.assertFalse(os.path.exists(fn))
|
|
|
|
|
|
class TestReplPartPowerIncrease(TestPartPowerIncrease, ReplProbeTest):
|
|
def test_main(self):
|
|
self._test_main()
|
|
|
|
def test_canceled(self):
|
|
self._test_main(cancel=True)
|
|
|
|
|
|
class TestECPartPowerIncrease(TestPartPowerIncrease, ECProbeTest):
|
|
def test_main(self):
|
|
self._test_main()
|
|
|
|
def test_canceled(self):
|
|
self._test_main(cancel=True)
|
|
|
|
|
|
if __name__ == '__main__':
|
|
main()
|