swift/test/unit/common/test_db_replicator.py

2265 lines
96 KiB
Python
Raw Normal View History

# Copyright (c) 2010-2012 OpenStack Foundation
2010-07-12 17:03:45 -05:00
#
# 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.
from __future__ import print_function
2010-07-12 17:03:45 -05:00
import unittest
from contextlib import contextmanager
import eventlet
2010-07-12 17:03:45 -05:00
import os
import logging
import errno
import math
import time
from shutil import rmtree, copy
from tempfile import mkdtemp, NamedTemporaryFile
import json
2010-07-12 17:03:45 -05:00
import mock
from mock import patch, call
from six.moves import reload_module
from swift.container.backend import DATADIR
2010-07-12 17:03:45 -05:00
from swift.common import db_replicator
from swift.common.utils import (normalize_timestamp, hash_path,
storage_directory, Timestamp)
from swift.common.exceptions import DriveNotMounted
from swift.common.swob import HTTPException
2010-07-12 17:03:45 -05:00
from test import unit
from test.unit import FakeLogger, attach_fake_replication_rpc
from test.unit.common.test_db import ExampleBroker
2010-07-12 17:03:45 -05:00
TEST_ACCOUNT_NAME = 'a c t'
TEST_CONTAINER_NAME = 'c o n'
2010-07-12 17:03:45 -05:00
def teardown_module():
"clean up my monkey patching"
reload_module(db_replicator)
2010-07-12 17:03:45 -05:00
2010-07-12 17:03:45 -05:00
@contextmanager
def lock_parent_directory(filename):
yield True
class FakeRing(object):
class Ring(object):
2010-07-12 17:03:45 -05:00
devs = []
def __init__(self, path, reload_time=15, ring_name=None):
2010-07-12 17:03:45 -05:00
pass
def get_part(self, account, container=None, obj=None):
return 0
2010-07-12 17:03:45 -05:00
def get_part_nodes(self, part):
return []
2010-07-12 17:03:45 -05:00
def get_more_nodes(self, *args):
return []
class FakeRingWithSingleNode(object):
class Ring(object):
devs = [dict(
id=1, weight=10.0, zone=1, ip='1.1.1.1', port=6200, device='sdb',
meta='', replication_ip='1.1.1.1', replication_port=6200, region=1
)]
def __init__(self, path, reload_time=15, ring_name=None):
pass
def get_part(self, account, container=None, obj=None):
return 0
def get_part_nodes(self, part):
return self.devs
def get_more_nodes(self, *args):
return (d for d in self.devs)
class FakeRingWithNodes(object):
class Ring(object):
devs = [dict(
id=1, weight=10.0, zone=1, ip='1.1.1.1', port=6200, device='sdb',
meta='', replication_ip='1.1.1.1', replication_port=6200, region=1
), dict(
id=2, weight=10.0, zone=2, ip='1.1.1.2', port=6200, device='sdb',
meta='', replication_ip='1.1.1.2', replication_port=6200, region=2
), dict(
id=3, weight=10.0, zone=3, ip='1.1.1.3', port=6200, device='sdb',
meta='', replication_ip='1.1.1.3', replication_port=6200, region=1
), dict(
id=4, weight=10.0, zone=4, ip='1.1.1.4', port=6200, device='sdb',
meta='', replication_ip='1.1.1.4', replication_port=6200, region=2
), dict(
id=5, weight=10.0, zone=5, ip='1.1.1.5', port=6200, device='sdb',
meta='', replication_ip='1.1.1.5', replication_port=6200, region=1
), dict(
id=6, weight=10.0, zone=6, ip='1.1.1.6', port=6200, device='sdb',
meta='', replication_ip='1.1.1.6', replication_port=6200, region=2
)]
def __init__(self, path, reload_time=15, ring_name=None):
pass
def get_part(self, account, container=None, obj=None):
return 0
def get_part_nodes(self, part):
return self.devs[:3]
def get_more_nodes(self, *args):
return (d for d in self.devs[3:])
class FakeProcess(object):
2010-07-12 17:03:45 -05:00
def __init__(self, *codes):
self.codes = iter(codes)
self.args = None
self.kwargs = None
2010-07-12 17:03:45 -05:00
def __call__(self, *args, **kwargs):
self.args = args
self.kwargs = kwargs
class Failure(object):
2010-07-12 17:03:45 -05:00
def communicate(innerself):
next_item = next(self.codes)
if isinstance(next_item, int):
innerself.returncode = next_item
return next_item
raise next_item
2010-07-12 17:03:45 -05:00
return Failure()
2010-07-12 17:03:45 -05:00
@contextmanager
def _mock_process(*args):
orig_process = db_replicator.subprocess.Popen
db_replicator.subprocess.Popen = FakeProcess(*args)
yield db_replicator.subprocess.Popen
2010-07-12 17:03:45 -05:00
db_replicator.subprocess.Popen = orig_process
class ReplHttp(object):
def __init__(self, response=None, set_status=200):
2010-07-12 17:03:45 -05:00
self.response = response
self.set_status = set_status
replicated = False
2010-07-12 17:03:45 -05:00
host = 'localhost'
node = {
'ip': '127.0.0.1',
'port': '6000',
'device': 'sdb',
}
def replicate(self, *args):
self.replicated = True
class Response(object):
status = self.set_status
2010-07-12 17:03:45 -05:00
data = self.response
2010-07-12 17:03:45 -05:00
def read(innerself):
return self.response
return Response()
class ChangingMtimesOs(object):
2010-07-12 17:03:45 -05:00
def __init__(self):
self.mtime = 0
def __call__(self, *args, **kwargs):
2010-07-12 17:03:45 -05:00
self.mtime += 1
return self.mtime
class FakeBroker(object):
2010-07-12 17:03:45 -05:00
db_file = __file__
get_repl_missing_table = False
stub_replication_info = None
db_type = 'container'
db_contains_type = 'object'
info = {'account': TEST_ACCOUNT_NAME, 'container': TEST_CONTAINER_NAME}
2010-07-12 17:03:45 -05:00
def __init__(self, *args, **kwargs):
self.locked = False
self.metadata = {}
2010-07-12 17:03:45 -05:00
return None
2010-07-12 17:03:45 -05:00
@contextmanager
def lock(self):
self.locked = True
2010-07-12 17:03:45 -05:00
yield True
self.locked = False
2010-07-12 17:03:45 -05:00
def get_sync(self, *args, **kwargs):
return 5
2010-07-12 17:03:45 -05:00
def get_syncs(self):
return []
2010-07-12 17:03:45 -05:00
def get_items_since(self, point, *args):
if point == 0:
return [{'ROWID': 1}]
if point == -1:
return [{'ROWID': 1}, {'ROWID': 2}]
2010-07-12 17:03:45 -05:00
return []
2010-07-12 17:03:45 -05:00
def merge_syncs(self, *args, **kwargs):
self.args = args
2010-07-12 17:03:45 -05:00
def merge_items(self, *args):
self.args = args
2010-07-12 17:03:45 -05:00
def get_replication_info(self):
if self.get_repl_missing_table:
raise Exception('no such table')
info = dict(self.info)
info.update({
'hash': 12345,
'delete_timestamp': 0,
'put_timestamp': 1,
'created_at': 1,
'count': 0,
'max_row': 99,
'id': 'ID',
'metadata': {}
})
if self.stub_replication_info:
info.update(self.stub_replication_info)
return info
def get_max_row(self, table=None):
return self.get_replication_info()['max_row']
def is_reclaimable(self, now, reclaim_age):
info = self.get_replication_info()
return info['count'] == 0 and (
(now - reclaim_age) >
info['delete_timestamp'] >
info['put_timestamp'])
def get_other_replication_items(self):
return None
2010-07-12 17:03:45 -05:00
def reclaim(self, item_timestamp, sync_timestamp):
pass
def newid(self, remote_d):
pass
def update_metadata(self, metadata):
self.metadata = metadata
def merge_timestamps(self, created_at, put_timestamp, delete_timestamp):
self.created_at = created_at
self.put_timestamp = put_timestamp
self.delete_timestamp = delete_timestamp
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
def get_brokers(self):
return [self]
class FakeAccountBroker(FakeBroker):
db_type = 'account'
db_contains_type = 'container'
info = {'account': TEST_ACCOUNT_NAME}
2010-07-12 17:03:45 -05:00
class TestReplicator(db_replicator.Replicator):
server_type = 'container'
ring_file = 'container.ring.gz'
brokerclass = FakeBroker
datadir = DATADIR
2010-07-12 17:03:45 -05:00
default_port = 1000
2010-07-12 17:03:45 -05:00
class TestDBReplicator(unittest.TestCase):
def setUp(self):
db_replicator.ring = FakeRing()
self.delete_db_calls = []
self._patchers = []
# recon cache path
self.recon_cache = mkdtemp()
rmtree(self.recon_cache, ignore_errors=1)
os.mkdir(self.recon_cache)
self.logger = unit.debug_logger('test-replicator')
def tearDown(self):
for patcher in self._patchers:
patcher.stop()
rmtree(self.recon_cache, ignore_errors=1)
def _patch(self, patching_fn, *args, **kwargs):
patcher = patching_fn(*args, **kwargs)
patched_thing = patcher.start()
self._patchers.append(patcher)
return patched_thing
def stub_delete_db(self, broker):
self.delete_db_calls.append('/path/to/file')
return True
def test_creation(self):
# later config should be extended to assert more config options
replicator = TestReplicator({'node_timeout': '3.5'})
self.assertEqual(replicator.node_timeout, 3.5)
2010-07-12 17:03:45 -05:00
def test_repl_connection(self):
node = {'replication_ip': '127.0.0.1', 'replication_port': 80,
'device': 'sdb1'}
2010-07-12 17:03:45 -05:00
conn = db_replicator.ReplConnection(node, '1234567890', 'abcdefg',
logging.getLogger())
2010-07-12 17:03:45 -05:00
def req(method, path, body, headers):
self.assertEqual(method, 'REPLICATE')
self.assertEqual(headers['Content-Type'], 'application/json')
class Resp(object):
def read(self):
return 'data'
2010-07-12 17:03:45 -05:00
resp = Resp()
conn.request = req
conn.getresponse = lambda *args: resp
self.assertEqual(conn.replicate(1, 2, 3), resp)
2010-07-12 17:03:45 -05:00
def other_req(method, path, body, headers):
raise Exception('blah')
conn.request = other_req
self.assertIsNone(conn.replicate(1, 2, 3))
2010-07-12 17:03:45 -05:00
def test_rsync_file(self):
2010-08-20 00:42:38 +00:00
replicator = TestReplicator({})
2010-07-12 17:03:45 -05:00
with _mock_process(-1):
self.assertEqual(
False,
replicator._rsync_file('/some/file', 'remote:/some/file'))
2010-07-12 17:03:45 -05:00
with _mock_process(0):
self.assertEqual(
True,
replicator._rsync_file('/some/file', 'remote:/some/file'))
2010-07-12 17:03:45 -05:00
def test_rsync_file_popen_args(self):
replicator = TestReplicator({})
with _mock_process(0) as process:
replicator._rsync_file('/some/file', 'remote:/some_file')
exp_args = ([
'rsync', '--quiet', '--no-motd',
'--timeout=%s' % int(math.ceil(replicator.node_timeout)),
'--contimeout=%s' % int(math.ceil(replicator.conn_timeout)),
'--whole-file', '/some/file', 'remote:/some_file'],)
self.assertEqual(exp_args, process.args)
def test_rsync_file_popen_args_whole_file_false(self):
replicator = TestReplicator({})
with _mock_process(0) as process:
replicator._rsync_file('/some/file', 'remote:/some_file', False)
exp_args = ([
'rsync', '--quiet', '--no-motd',
'--timeout=%s' % int(math.ceil(replicator.node_timeout)),
'--contimeout=%s' % int(math.ceil(replicator.conn_timeout)),
'/some/file', 'remote:/some_file'],)
self.assertEqual(exp_args, process.args)
def test_rsync_file_popen_args_different_region_and_rsync_compress(self):
replicator = TestReplicator({})
for rsync_compress in (False, True):
replicator.rsync_compress = rsync_compress
for different_region in (False, True):
with _mock_process(0) as process:
replicator._rsync_file('/some/file', 'remote:/some_file',
False, different_region)
if rsync_compress and different_region:
# --compress arg should be passed to rsync binary
# only when rsync_compress option is enabled
# AND destination node is in a different
# region
self.assertTrue('--compress' in process.args[0])
else:
self.assertFalse('--compress' in process.args[0])
2010-07-12 17:03:45 -05:00
def test_rsync_db(self):
2010-08-20 00:42:38 +00:00
replicator = TestReplicator({})
replicator._rsync_file = lambda *args, **kwargs: True
fake_device = {'replication_ip': '127.0.0.1', 'device': 'sda1'}
replicator._rsync_db(FakeBroker(), fake_device, ReplHttp(), 'abcd')
2010-07-12 17:03:45 -05:00
def test_rsync_db_rsync_file_call(self):
fake_device = {'ip': '127.0.0.1', 'port': '0',
'replication_ip': '127.0.0.1', 'replication_port': '0',
'device': 'sda1'}
class MyTestReplicator(TestReplicator):
def __init__(self, db_file, remote_file):
super(MyTestReplicator, self).__init__({})
self.db_file = db_file
self.remote_file = remote_file
def _rsync_file(self_, db_file, remote_file, whole_file=True,
different_region=False):
self.assertEqual(self_.db_file, db_file)
self.assertEqual(self_.remote_file, remote_file)
self_._rsync_file_called = True
return False
Allows to configure the rsync modules where the replicators will send data Currently, the rsync module where the replicators send data is static. It forbids administrators to set rsync configuration based on their current deployment or needs. As an example, the rsyncd configuration example encourages to set a connections limit for the modules account, container and object. It permits to protect devices from excessives parallels connections, because it would impact performances. On a server with many devices, it is tempting to increase this number proportionally, but nothing guarantees that the distribution of the connections will be balanced. In the worst scenario, a single device can receive all the connections, which is a severe impact on performances. This commit adds a new option named 'rsync_module' to the *-replicator sections of the *-server configuration file. This configuration variable can be extrapolated with device attributes like ip, port, device, zone, ... by using the format {NAME}. eg: rsync_module = {replication_ip}::object_{device} With this configuration, an administrators can solve the problem of connections distribution by creating one module per device in rsyncd configuration. The default values are backward compatible: {replication_ip}::account {replication_ip}::container {replication_ip}::object Option vm_test_mode is deprecated by this commit, but backward compatibility is maintained. The option is only effective when rsync_module is not set. In that case, {replication_port} is appended to the default value of rsync_module. Change-Id: Iad91df50dadbe96c921181797799b4444323ce2e
2015-06-16 12:47:26 +02:00
broker = FakeBroker()
remote_file = '127.0.0.1::container/sda1/tmp/abcd'
replicator = MyTestReplicator(broker.db_file, remote_file)
replicator._rsync_db(broker, fake_device, ReplHttp(), 'abcd')
self.assertTrue(replicator._rsync_file_called)
def test_rsync_db_rsync_file_failure(self):
class MyTestReplicator(TestReplicator):
def __init__(self):
super(MyTestReplicator, self).__init__({})
self._rsync_file_called = False
def _rsync_file(self_, *args, **kwargs):
self.assertEqual(
False, self_._rsync_file_called,
'_sync_file() should only be called once')
self_._rsync_file_called = True
return False
with patch('os.path.exists', lambda *args: True):
replicator = MyTestReplicator()
fake_device = {'ip': '127.0.0.1', 'replication_ip': '127.0.0.1',
'device': 'sda1'}
replicator._rsync_db(FakeBroker(), fake_device, ReplHttp(), 'abcd')
self.assertEqual(True, replicator._rsync_file_called)
def test_rsync_db_change_after_sync(self):
class MyTestReplicator(TestReplicator):
def __init__(self, broker):
super(MyTestReplicator, self).__init__({})
self.broker = broker
self._rsync_file_call_count = 0
def _rsync_file(self_, db_file, remote_file, whole_file=True,
different_region=False):
self_._rsync_file_call_count += 1
if self_._rsync_file_call_count == 1:
self.assertEqual(True, whole_file)
self.assertEqual(False, self_.broker.locked)
elif self_._rsync_file_call_count == 2:
self.assertEqual(False, whole_file)
self.assertEqual(True, self_.broker.locked)
else:
raise RuntimeError('_rsync_file() called too many times')
return True
# with journal file
with patch('os.path.exists', lambda *args: True):
broker = FakeBroker()
replicator = MyTestReplicator(broker)
fake_device = {'ip': '127.0.0.1', 'replication_ip': '127.0.0.1',
'device': 'sda1'}
replicator._rsync_db(broker, fake_device, ReplHttp(), 'abcd')
self.assertEqual(2, replicator._rsync_file_call_count)
# with new mtime
with patch('os.path.exists', lambda *args: False):
with patch('os.path.getmtime', ChangingMtimesOs()):
broker = FakeBroker()
replicator = MyTestReplicator(broker)
fake_device = {'ip': '127.0.0.1',
'replication_ip': '127.0.0.1',
'device': 'sda1'}
replicator._rsync_db(broker, fake_device, ReplHttp(), 'abcd')
self.assertEqual(2, replicator._rsync_file_call_count)
2010-07-12 17:03:45 -05:00
def test_in_sync(self):
2010-08-20 00:42:38 +00:00
replicator = TestReplicator({})
self.assertEqual(replicator._in_sync(
{'id': 'a', 'point': 0, 'max_row': 0, 'hash': 'b'},
2010-07-12 17:03:45 -05:00
{'id': 'a', 'point': -1, 'max_row': 0, 'hash': 'b'},
FakeBroker(), -1), True)
self.assertEqual(replicator._in_sync(
2010-07-12 17:03:45 -05:00
{'id': 'a', 'point': -1, 'max_row': 0, 'hash': 'b'},
{'id': 'a', 'point': -1, 'max_row': 10, 'hash': 'b'},
FakeBroker(), -1), True)
self.assertEqual(bool(replicator._in_sync(
2010-07-12 17:03:45 -05:00
{'id': 'a', 'point': -1, 'max_row': 0, 'hash': 'c'},
{'id': 'a', 'point': -1, 'max_row': 10, 'hash': 'd'},
FakeBroker(), -1)), False)
def test_run_once_no_local_device_in_ring(self):
logger = unit.debug_logger('test-replicator')
replicator = TestReplicator({'recon_cache_path': self.recon_cache},
logger=logger)
with patch('swift.common.db_replicator.whataremyips',
return_value=['127.0.0.1']):
replicator.run_once()
expected = [
"Can't find itself 127.0.0.1 with port 1000 "
"in ring file, not replicating",
]
self.assertEqual(expected, logger.get_lines_for_level('error'))
def test_run_once_with_local_device_in_ring(self):
logger = unit.debug_logger('test-replicator')
base = 'swift.common.db_replicator.'
with patch(base + 'whataremyips', return_value=['1.1.1.1']), \
patch(base + 'ring', FakeRingWithNodes()):
replicator = TestReplicator({'bind_port': 6200,
'recon_cache_path': self.recon_cache},
logger=logger)
replicator.run_once()
self.assertFalse(logger.get_lines_for_level('error'))
2010-07-12 17:03:45 -05:00
def test_run_once_no_ips(self):
replicator = TestReplicator({}, logger=unit.FakeLogger())
self._patch(patch.object, db_replicator, 'whataremyips',
Allow 1+ object-servers-per-disk deployment Enabled by a new > 0 integer config value, "servers_per_port" in the [DEFAULT] config section for object-server and/or replication server configs. The setting's integer value determines how many different object-server workers handle requests for any single unique local port in the ring. In this mode, the parent swift-object-server process continues to run as the original user (i.e. root if low-port binding is required), binds to all ports as defined in the ring, and forks off the specified number of workers per listen socket. The child, per-port servers drop privileges and behave pretty much how object-server workers always have, except that because the ring has unique ports per disk, the object-servers will only be handling requests for a single disk. The parent process detects dead servers and restarts them (with the correct listen socket), starts missing servers when an updated ring file is found with a device on the server with a new port, and kills extraneous servers when their port is found to no longer be in the ring. The ring files are stat'ed at most every "ring_check_interval" seconds, as configured in the object-server config (same default of 15s). Immediately stopping all swift-object-worker processes still works by sending the parent a SIGTERM. Likewise, a SIGHUP to the parent process still causes the parent process to close all listen sockets and exit, allowing existing children to finish serving their existing requests. The drop_privileges helper function now has an optional param to suppress the setsid() call, which otherwise screws up the child workers' process management. The class method RingData.load() can be told to only load the ring metadata (i.e. everything except replica2part2dev_id) with the optional kwarg, header_only=True. This is used to keep the parent and all forked off workers from unnecessarily having full copies of all storage policy rings in memory. A new helper class, swift.common.storage_policy.BindPortsCache, provides a method to return a set of all device ports in all rings for the server on which it is instantiated (identified by its set of IP addresses). The BindPortsCache instance will track mtimes of ring files, so they are not opened more frequently than necessary. This patch includes enhancements to the probe tests and object-replicator/object-reconstructor config plumbing to allow the probe tests to work correctly both in the "normal" config (same IP but unique ports for each SAIO "server") and a server-per-port setup where each SAIO "server" must have a unique IP address and unique port per disk within each "server". The main probe tests only work with 4 servers and 4 disks, but you can see the difference in the rings for the EC probe tests where there are 2 disks per server for a total of 8 disks. Specifically, swift.common.ring.utils.is_local_device() will ignore the ports when the "my_port" argument is None. Then, object-replicator and object-reconstructor both set self.bind_port to None if server_per_port is enabled. Bonus improvement for IPv6 addresses in is_local_device(). This PR for vagrant-swift-all-in-one will aid in testing this patch: https://github.com/swiftstack/vagrant-swift-all-in-one/pull/16/ Also allow SAIO to answer is_local_device() better; common SAIO setups have multiple "servers" all on the same host with different ports for the different "servers" (which happen to match the IPs specified in the rings for the devices on each of those "servers"). However, you can configure the SAIO to have different localhost IP addresses (e.g. 127.0.0.1, 127.0.0.2, etc.) in the ring and in the servers' config files' bind_ip setting. This new whataremyips() implementation combined with a little plumbing allows is_local_device() to accurately answer, even on an SAIO. In the default case (an unspecified bind_ip defaults to '0.0.0.0') as well as an explict "bind to everything" like '0.0.0.0' or '::', whataremyips() behaves as it always has, returning all IP addresses for the server. Also updated probe tests to handle each "server" in the SAIO having a unique IP address. For some (noisy) benchmarks that show servers_per_port=X is at least as good as the same number of "normal" workers: https://gist.github.com/dbishop/c214f89ca708a6b1624a#file-summary-md Benchmarks showing the benefits of I/O isolation with a small number of slow disks: https://gist.github.com/dbishop/fd0ab067babdecfb07ca#file-results-md If you were wondering what the overhead of threads_per_disk looks like: https://gist.github.com/dbishop/1d14755fedc86a161718#file-tabular_results-md DocImpact Change-Id: I2239a4000b41a7e7cc53465ce794af49d44796c6
2015-05-14 22:14:15 -07:00
lambda *a, **kw: [])
replicator.run_once()
self.assertEqual(
replicator.logger.log_dict['error'],
[(('ERROR Failed to get my own IPs?',), {})])
def test_run_once_node_is_not_mounted(self):
db_replicator.ring = FakeRingWithSingleNode()
Allow 1+ object-servers-per-disk deployment Enabled by a new > 0 integer config value, "servers_per_port" in the [DEFAULT] config section for object-server and/or replication server configs. The setting's integer value determines how many different object-server workers handle requests for any single unique local port in the ring. In this mode, the parent swift-object-server process continues to run as the original user (i.e. root if low-port binding is required), binds to all ports as defined in the ring, and forks off the specified number of workers per listen socket. The child, per-port servers drop privileges and behave pretty much how object-server workers always have, except that because the ring has unique ports per disk, the object-servers will only be handling requests for a single disk. The parent process detects dead servers and restarts them (with the correct listen socket), starts missing servers when an updated ring file is found with a device on the server with a new port, and kills extraneous servers when their port is found to no longer be in the ring. The ring files are stat'ed at most every "ring_check_interval" seconds, as configured in the object-server config (same default of 15s). Immediately stopping all swift-object-worker processes still works by sending the parent a SIGTERM. Likewise, a SIGHUP to the parent process still causes the parent process to close all listen sockets and exit, allowing existing children to finish serving their existing requests. The drop_privileges helper function now has an optional param to suppress the setsid() call, which otherwise screws up the child workers' process management. The class method RingData.load() can be told to only load the ring metadata (i.e. everything except replica2part2dev_id) with the optional kwarg, header_only=True. This is used to keep the parent and all forked off workers from unnecessarily having full copies of all storage policy rings in memory. A new helper class, swift.common.storage_policy.BindPortsCache, provides a method to return a set of all device ports in all rings for the server on which it is instantiated (identified by its set of IP addresses). The BindPortsCache instance will track mtimes of ring files, so they are not opened more frequently than necessary. This patch includes enhancements to the probe tests and object-replicator/object-reconstructor config plumbing to allow the probe tests to work correctly both in the "normal" config (same IP but unique ports for each SAIO "server") and a server-per-port setup where each SAIO "server" must have a unique IP address and unique port per disk within each "server". The main probe tests only work with 4 servers and 4 disks, but you can see the difference in the rings for the EC probe tests where there are 2 disks per server for a total of 8 disks. Specifically, swift.common.ring.utils.is_local_device() will ignore the ports when the "my_port" argument is None. Then, object-replicator and object-reconstructor both set self.bind_port to None if server_per_port is enabled. Bonus improvement for IPv6 addresses in is_local_device(). This PR for vagrant-swift-all-in-one will aid in testing this patch: https://github.com/swiftstack/vagrant-swift-all-in-one/pull/16/ Also allow SAIO to answer is_local_device() better; common SAIO setups have multiple "servers" all on the same host with different ports for the different "servers" (which happen to match the IPs specified in the rings for the devices on each of those "servers"). However, you can configure the SAIO to have different localhost IP addresses (e.g. 127.0.0.1, 127.0.0.2, etc.) in the ring and in the servers' config files' bind_ip setting. This new whataremyips() implementation combined with a little plumbing allows is_local_device() to accurately answer, even on an SAIO. In the default case (an unspecified bind_ip defaults to '0.0.0.0') as well as an explict "bind to everything" like '0.0.0.0' or '::', whataremyips() behaves as it always has, returning all IP addresses for the server. Also updated probe tests to handle each "server" in the SAIO having a unique IP address. For some (noisy) benchmarks that show servers_per_port=X is at least as good as the same number of "normal" workers: https://gist.github.com/dbishop/c214f89ca708a6b1624a#file-summary-md Benchmarks showing the benefits of I/O isolation with a small number of slow disks: https://gist.github.com/dbishop/fd0ab067babdecfb07ca#file-results-md If you were wondering what the overhead of threads_per_disk looks like: https://gist.github.com/dbishop/1d14755fedc86a161718#file-tabular_results-md DocImpact Change-Id: I2239a4000b41a7e7cc53465ce794af49d44796c6
2015-05-14 22:14:15 -07:00
# If a bind_ip is specified, it's plumbed into whataremyips() and
# returned by itself.
conf = {'mount_check': 'true', 'bind_ip': '1.1.1.1',
'bind_port': 6200}
replicator = TestReplicator(conf, logger=unit.FakeLogger())
self.assertEqual(replicator.mount_check, True)
self.assertEqual(replicator.port, 6200)
err = ValueError('Boom!')
def mock_check_drive(root, device, mount_check):
self.assertEqual(root, replicator.root)
self.assertEqual(device, replicator.ring.devs[0]['device'])
self.assertEqual(mount_check, True)
raise err
self._patch(patch.object, db_replicator, 'check_drive',
mock_check_drive)
replicator.run_once()
self.assertEqual(
replicator.logger.log_dict['warning'],
[(('Skipping: %s', (err, )), {})])
def test_run_once_node_is_mounted(self):
db_replicator.ring = FakeRingWithSingleNode()
conf = {'mount_check': 'true', 'bind_port': 6200}
replicator = TestReplicator(conf, logger=unit.FakeLogger())
self.assertEqual(replicator.mount_check, True)
self.assertEqual(replicator.port, 6200)
def mock_unlink_older_than(path, mtime):
self.assertEqual(path,
os.path.join(replicator.root,
replicator.ring.devs[0]['device'],
'tmp'))
self.assertTrue(time.time() - replicator.reclaim_age >= mtime)
def mock_spawn_n(fn, part, object_file, node_id):
self.assertEqual('123', part)
self.assertEqual('/srv/node/sda/c.db', object_file)
self.assertEqual(1, node_id)
self._patch(patch.object, db_replicator, 'whataremyips',
Allow 1+ object-servers-per-disk deployment Enabled by a new > 0 integer config value, "servers_per_port" in the [DEFAULT] config section for object-server and/or replication server configs. The setting's integer value determines how many different object-server workers handle requests for any single unique local port in the ring. In this mode, the parent swift-object-server process continues to run as the original user (i.e. root if low-port binding is required), binds to all ports as defined in the ring, and forks off the specified number of workers per listen socket. The child, per-port servers drop privileges and behave pretty much how object-server workers always have, except that because the ring has unique ports per disk, the object-servers will only be handling requests for a single disk. The parent process detects dead servers and restarts them (with the correct listen socket), starts missing servers when an updated ring file is found with a device on the server with a new port, and kills extraneous servers when their port is found to no longer be in the ring. The ring files are stat'ed at most every "ring_check_interval" seconds, as configured in the object-server config (same default of 15s). Immediately stopping all swift-object-worker processes still works by sending the parent a SIGTERM. Likewise, a SIGHUP to the parent process still causes the parent process to close all listen sockets and exit, allowing existing children to finish serving their existing requests. The drop_privileges helper function now has an optional param to suppress the setsid() call, which otherwise screws up the child workers' process management. The class method RingData.load() can be told to only load the ring metadata (i.e. everything except replica2part2dev_id) with the optional kwarg, header_only=True. This is used to keep the parent and all forked off workers from unnecessarily having full copies of all storage policy rings in memory. A new helper class, swift.common.storage_policy.BindPortsCache, provides a method to return a set of all device ports in all rings for the server on which it is instantiated (identified by its set of IP addresses). The BindPortsCache instance will track mtimes of ring files, so they are not opened more frequently than necessary. This patch includes enhancements to the probe tests and object-replicator/object-reconstructor config plumbing to allow the probe tests to work correctly both in the "normal" config (same IP but unique ports for each SAIO "server") and a server-per-port setup where each SAIO "server" must have a unique IP address and unique port per disk within each "server". The main probe tests only work with 4 servers and 4 disks, but you can see the difference in the rings for the EC probe tests where there are 2 disks per server for a total of 8 disks. Specifically, swift.common.ring.utils.is_local_device() will ignore the ports when the "my_port" argument is None. Then, object-replicator and object-reconstructor both set self.bind_port to None if server_per_port is enabled. Bonus improvement for IPv6 addresses in is_local_device(). This PR for vagrant-swift-all-in-one will aid in testing this patch: https://github.com/swiftstack/vagrant-swift-all-in-one/pull/16/ Also allow SAIO to answer is_local_device() better; common SAIO setups have multiple "servers" all on the same host with different ports for the different "servers" (which happen to match the IPs specified in the rings for the devices on each of those "servers"). However, you can configure the SAIO to have different localhost IP addresses (e.g. 127.0.0.1, 127.0.0.2, etc.) in the ring and in the servers' config files' bind_ip setting. This new whataremyips() implementation combined with a little plumbing allows is_local_device() to accurately answer, even on an SAIO. In the default case (an unspecified bind_ip defaults to '0.0.0.0') as well as an explict "bind to everything" like '0.0.0.0' or '::', whataremyips() behaves as it always has, returning all IP addresses for the server. Also updated probe tests to handle each "server" in the SAIO having a unique IP address. For some (noisy) benchmarks that show servers_per_port=X is at least as good as the same number of "normal" workers: https://gist.github.com/dbishop/c214f89ca708a6b1624a#file-summary-md Benchmarks showing the benefits of I/O isolation with a small number of slow disks: https://gist.github.com/dbishop/fd0ab067babdecfb07ca#file-results-md If you were wondering what the overhead of threads_per_disk looks like: https://gist.github.com/dbishop/1d14755fedc86a161718#file-tabular_results-md DocImpact Change-Id: I2239a4000b41a7e7cc53465ce794af49d44796c6
2015-05-14 22:14:15 -07:00
lambda *a, **kw: ['1.1.1.1'])
self._patch(patch.object, db_replicator, 'unlink_older_than',
mock_unlink_older_than)
self._patch(patch.object, db_replicator, 'roundrobin_datadirs',
lambda *args: [('123', '/srv/node/sda/c.db', 1)])
self._patch(patch.object, replicator.cpool, 'spawn_n', mock_spawn_n)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(ismount=True) as mocks:
mock_os.path.isdir.return_value = True
replicator.run_once()
mock_os.path.isdir.assert_called_with(
os.path.join(replicator.root,
replicator.ring.devs[0]['device'],
replicator.datadir))
self.assertEqual([
mock.call(os.path.join(
replicator.root,
replicator.ring.devs[0]['device'])),
], mocks['ismount'].call_args_list)
2010-07-12 17:03:45 -05:00
def test_usync(self):
fake_http = ReplHttp()
2010-08-20 00:42:38 +00:00
replicator = TestReplicator({})
2010-07-12 17:03:45 -05:00
replicator._usync_db(0, FakeBroker(), fake_http, '12345', '67890')
def test_usync_http_error_above_300(self):
fake_http = ReplHttp(set_status=301)
replicator = TestReplicator({})
self.assertFalse(
replicator._usync_db(0, FakeBroker(), fake_http, '12345', '67890'))
def test_usync_http_error_below_200(self):
fake_http = ReplHttp(set_status=101)
replicator = TestReplicator({})
self.assertFalse(
replicator._usync_db(0, FakeBroker(), fake_http, '12345', '67890'))
@mock.patch('swift.common.db_replicator.dump_recon_cache')
@mock.patch('swift.common.db_replicator.time.time', return_value=1234.5678)
def test_stats(self, mock_time, mock_recon_cache):
logger = unit.debug_logger('test-replicator')
replicator = TestReplicator({}, logger=logger)
2010-07-12 17:03:45 -05:00
replicator._zero_stats()
self.assertEqual(replicator.stats['start'], mock_time.return_value)
2010-07-12 17:03:45 -05:00
replicator._report_stats()
self.assertEqual(logger.get_lines_for_level('info'), [
'Attempted to replicate 0 dbs in 0.00000 seconds (0.00000/s)',
'Removed 0 dbs',
'0 successes, 0 failures',
'diff:0 diff_capped:0 empty:0 hashmatch:0 no_change:0 '
'remote_merge:0 rsync:0 ts_repl:0',
])
self.assertEqual(1, len(mock_recon_cache.mock_calls))
self.assertEqual(mock_recon_cache.mock_calls[0][1][0], {
'replication_time': 0.0,
'replication_last': mock_time.return_value,
'replication_stats': replicator.stats,
})
mock_recon_cache.reset_mock()
logger.clear()
replicator.stats.update({
'attempted': 30,
'success': 25,
'remove': 9,
'failure': 1,
'diff': 5,
'diff_capped': 4,
'empty': 7,
'hashmatch': 8,
'no_change': 6,
'remote_merge': 2,
'rsync': 3,
'ts_repl': 10,
})
mock_time.return_value += 246.813576
replicator._report_stats()
self.maxDiff = None
self.assertEqual(logger.get_lines_for_level('info'), [
'Attempted to replicate 30 dbs in 246.81358 seconds (0.12155/s)',
'Removed 9 dbs',
'25 successes, 1 failures',
'diff:5 diff_capped:4 empty:7 hashmatch:8 no_change:6 '
'remote_merge:2 rsync:3 ts_repl:10',
])
self.assertEqual(1, len(mock_recon_cache.mock_calls))
self.assertEqual(mock_recon_cache.mock_calls[0][1][0], {
'replication_time': 246.813576,
'replication_last': mock_time.return_value,
'replication_stats': replicator.stats,
})
2010-07-12 17:03:45 -05:00
def test_replicate_object(self):
# verify return values from replicate_object
db_replicator.ring = FakeRingWithNodes()
db_path = '/path/to/file'
replicator = TestReplicator({}, logger=FakeLogger())
info = FakeBroker().get_replication_info()
# make remote appear to be in sync
rinfo = {'point': info['max_row'], 'id': 'remote_id'}
class FakeResponse(object):
def __init__(self, status, rinfo):
self._status = status
self.data = json.dumps(rinfo)
@property
def status(self):
if isinstance(self._status, (Exception, eventlet.Timeout)):
raise self._status
return self._status
# all requests fail
replicate = 'swift.common.db_replicator.ReplConnection.replicate'
with mock.patch(replicate) as fake_replicate:
fake_replicate.side_effect = [
FakeResponse(500, None),
FakeResponse(500, None),
FakeResponse(500, None)]
with mock.patch.object(replicator, 'delete_db') as mock_delete:
res = replicator._replicate_object('0', db_path, 'node_id')
self.assertRaises(StopIteration, next, fake_replicate.side_effect)
self.assertEqual((False, [False, False, False]), res)
self.assertEqual(0, mock_delete.call_count)
self.assertFalse(replicator.logger.get_lines_for_level('error'))
self.assertFalse(replicator.logger.get_lines_for_level('warning'))
replicator.logger.clear()
with mock.patch(replicate) as fake_replicate:
fake_replicate.side_effect = [
FakeResponse(Exception('ugh'), None),
FakeResponse(eventlet.Timeout(), None),
FakeResponse(200, rinfo)]
with mock.patch.object(replicator, 'delete_db') as mock_delete:
res = replicator._replicate_object('0', db_path, 'node_id')
self.assertRaises(StopIteration, next, fake_replicate.side_effect)
self.assertEqual((False, [False, False, True]), res)
self.assertEqual(0, mock_delete.call_count)
lines = replicator.logger.get_lines_for_level('error')
self.assertIn('ERROR syncing', lines[0])
self.assertIn('ERROR syncing', lines[1])
self.assertFalse(lines[2:])
self.assertFalse(replicator.logger.get_lines_for_level('warning'))
replicator.logger.clear()
# partial success
with mock.patch(replicate) as fake_replicate:
fake_replicate.side_effect = [
FakeResponse(200, rinfo),
FakeResponse(200, rinfo),
FakeResponse(500, None)]
with mock.patch.object(replicator, 'delete_db') as mock_delete:
res = replicator._replicate_object('0', db_path, 'node_id')
self.assertRaises(StopIteration, next, fake_replicate.side_effect)
self.assertEqual((False, [True, True, False]), res)
self.assertEqual(0, mock_delete.call_count)
self.assertFalse(replicator.logger.get_lines_for_level('error'))
self.assertFalse(replicator.logger.get_lines_for_level('warning'))
replicator.logger.clear()
# 507 triggers additional requests
with mock.patch(replicate) as fake_replicate:
fake_replicate.side_effect = [
FakeResponse(200, rinfo),
FakeResponse(200, rinfo),
FakeResponse(507, None),
FakeResponse(507, None),
FakeResponse(200, rinfo)]
with mock.patch.object(replicator, 'delete_db') as mock_delete:
res = replicator._replicate_object('0', db_path, 'node_id')
self.assertRaises(StopIteration, next, fake_replicate.side_effect)
self.assertEqual((False, [True, True, False, False, True]), res)
self.assertEqual(0, mock_delete.call_count)
lines = replicator.logger.get_lines_for_level('error')
self.assertIn('Remote drive not mounted', lines[0])
self.assertIn('Remote drive not mounted', lines[1])
self.assertFalse(lines[2:])
self.assertFalse(replicator.logger.get_lines_for_level('warning'))
replicator.logger.clear()
# all requests succeed; node id == 'node_id' causes node to be
# considered a handoff so expect the db to be deleted
with mock.patch(replicate) as fake_replicate:
fake_replicate.side_effect = [
FakeResponse(200, rinfo),
FakeResponse(200, rinfo),
FakeResponse(200, rinfo)]
with mock.patch.object(replicator, 'delete_db') as mock_delete:
res = replicator._replicate_object('0', db_path, 'node_id')
self.assertRaises(StopIteration, next, fake_replicate.side_effect)
self.assertEqual((True, [True, True, True]), res)
self.assertEqual(1, mock_delete.call_count)
self.assertFalse(replicator.logger.get_lines_for_level('error'))
self.assertFalse(replicator.logger.get_lines_for_level('warning'))
2010-07-12 17:03:45 -05:00
def test_replicate_object_quarantine(self):
replicator = TestReplicator({})
self._patch(patch.object, replicator.brokerclass, 'db_file',
'/a/b/c/d/e/hey')
self._patch(patch.object, replicator.brokerclass,
'get_repl_missing_table', True)
def mock_renamer(was, new, fsync=False, cause_colision=False):
if cause_colision and '-' not in new:
raise OSError(errno.EEXIST, "File already exists")
self.assertEqual('/a/b/c/d/e', was)
if '-' in new:
self.assertTrue(
new.startswith('/a/quarantined/containers/e-'))
else:
self.assertEqual('/a/quarantined/containers/e', new)
def mock_renamer_error(was, new, fsync):
return mock_renamer(was, new, fsync, cause_colision=True)
with patch.object(db_replicator, 'renamer', mock_renamer):
replicator._replicate_object('0', 'file', 'node_id')
# try the double quarantine
with patch.object(db_replicator, 'renamer', mock_renamer_error):
replicator._replicate_object('0', 'file', 'node_id')
2010-07-12 17:03:45 -05:00
def test_replicate_object_delete_because_deleted(self):
replicator = TestReplicator({})
try:
replicator.delete_db = self.stub_delete_db
replicator.brokerclass.stub_replication_info = {
'delete_timestamp': 2, 'put_timestamp': 1}
replicator._replicate_object('0', '/path/to/file', 'node_id')
finally:
replicator.brokerclass.stub_replication_info = None
self.assertEqual(['/path/to/file'], self.delete_db_calls)
def test_replicate_object_delete_because_not_shouldbehere(self):
replicator = TestReplicator({})
replicator.ring = FakeRingWithNodes().Ring('path')
replicator.brokerclass = FakeAccountBroker
replicator._repl_to_node = lambda *args: True
replicator.delete_db = self.stub_delete_db
orig_cleanup = replicator.cleanup_post_replicate
with mock.patch.object(replicator, 'cleanup_post_replicate',
side_effect=orig_cleanup) as mock_cleanup:
replicator._replicate_object('0', '/path/to/file', 'node_id')
mock_cleanup.assert_called_once_with(mock.ANY, mock.ANY, [True] * 3)
self.assertIsInstance(mock_cleanup.call_args[0][0],
replicator.brokerclass)
self.assertEqual(['/path/to/file'], self.delete_db_calls)
self.assertEqual(0, replicator.stats['failure'])
def test_replicate_object_delete_delegated_to_cleanup_post_replicate(self):
replicator = TestReplicator({})
replicator.ring = FakeRingWithNodes().Ring('path')
replicator.brokerclass = FakeAccountBroker
replicator._repl_to_node = lambda *args: True
replicator.delete_db = self.stub_delete_db
# cleanup succeeds
with mock.patch.object(replicator, 'cleanup_post_replicate',
return_value=True) as mock_cleanup:
replicator._replicate_object('0', '/path/to/file', 'node_id')
mock_cleanup.assert_called_once_with(mock.ANY, mock.ANY, [True] * 3)
self.assertIsInstance(mock_cleanup.call_args[0][0],
replicator.brokerclass)
self.assertFalse(self.delete_db_calls)
self.assertEqual(0, replicator.stats['failure'])
self.assertEqual(3, replicator.stats['success'])
# cleanup fails
replicator._zero_stats()
with mock.patch.object(replicator, 'cleanup_post_replicate',
return_value=False) as mock_cleanup:
replicator._replicate_object('0', '/path/to/file', 'node_id')
mock_cleanup.assert_called_once_with(mock.ANY, mock.ANY, [True] * 3)
self.assertIsInstance(mock_cleanup.call_args[0][0],
replicator.brokerclass)
self.assertFalse(self.delete_db_calls)
self.assertEqual(3, replicator.stats['failure'])
self.assertEqual(0, replicator.stats['success'])
# shouldbehere True - cleanup not required
replicator._zero_stats()
primary_node_id = replicator.ring.get_part_nodes('0')[0]['id']
with mock.patch.object(replicator, 'cleanup_post_replicate',
return_value=True) as mock_cleanup:
replicator._replicate_object('0', '/path/to/file', primary_node_id)
mock_cleanup.assert_not_called()
self.assertFalse(self.delete_db_calls)
self.assertEqual(0, replicator.stats['failure'])
self.assertEqual(2, replicator.stats['success'])
def test_cleanup_post_replicate(self):
replicator = TestReplicator({}, logger=self.logger)
replicator.ring = FakeRingWithNodes().Ring('path')
broker = FakeBroker()
replicator._repl_to_node = lambda *args: True
info = broker.get_replication_info()
with mock.patch.object(replicator, 'delete_db') as mock_delete_db:
res = replicator.cleanup_post_replicate(
broker, info, [False] * 3)
mock_delete_db.assert_not_called()
self.assertTrue(res)
self.assertEqual(['Not deleting db %s (0/3 success)' % broker.db_file],
replicator.logger.get_lines_for_level('debug'))
replicator.logger.clear()
with mock.patch.object(replicator, 'delete_db') as mock_delete_db:
res = replicator.cleanup_post_replicate(
broker, info, [True, False, True])
mock_delete_db.assert_not_called()
self.assertTrue(res)
self.assertEqual(['Not deleting db %s (2/3 success)' % broker.db_file],
replicator.logger.get_lines_for_level('debug'))
replicator.logger.clear()
broker.stub_replication_info = {'max_row': 101}
with mock.patch.object(replicator, 'delete_db') as mock_delete_db:
res = replicator.cleanup_post_replicate(
broker, info, [True] * 3)
mock_delete_db.assert_not_called()
self.assertTrue(res)
self.assertEqual(['Not deleting db %s (2 new rows)' % broker.db_file],
replicator.logger.get_lines_for_level('debug'))
replicator.logger.clear()
broker.stub_replication_info = {'max_row': 98}
with mock.patch.object(replicator, 'delete_db') as mock_delete_db:
res = replicator.cleanup_post_replicate(
broker, info, [True] * 3)
mock_delete_db.assert_not_called()
self.assertTrue(res)
broker.stub_replication_info = None
self.assertEqual(['Not deleting db %s (negative max_row_delta: -1)' %
broker.db_file],
replicator.logger.get_lines_for_level('error'))
replicator.logger.clear()
with mock.patch.object(replicator, 'delete_db') as mock_delete_db:
res = replicator.cleanup_post_replicate(
broker, info, [True] * 3)
mock_delete_db.assert_called_once_with(broker)
self.assertTrue(res)
self.assertEqual(['Successfully deleted db %s' % broker.db_file],
replicator.logger.get_lines_for_level('debug'))
replicator.logger.clear()
with mock.patch.object(replicator, 'delete_db',
return_value=False) as mock_delete_db:
res = replicator.cleanup_post_replicate(
broker, info, [True] * 3)
mock_delete_db.assert_called_once_with(broker)
self.assertFalse(res)
self.assertEqual(['Failed to delete db %s' % broker.db_file],
replicator.logger.get_lines_for_level('debug'))
replicator.logger.clear()
def test_replicate_object_with_exception(self):
replicator = TestReplicator({})
replicator.ring = FakeRingWithNodes().Ring('path')
replicator.brokerclass = FakeAccountBroker
replicator.delete_db = self.stub_delete_db
replicator._repl_to_node = mock.Mock(side_effect=Exception())
replicator._replicate_object('0', '/path/to/file',
replicator.ring.devs[0]['id'])
self.assertEqual(2, replicator._repl_to_node.call_count)
# with one DriveNotMounted exception called on +1 more replica
replicator._repl_to_node = mock.Mock(side_effect=[DriveNotMounted()])
replicator._replicate_object('0', '/path/to/file',
replicator.ring.devs[0]['id'])
self.assertEqual(3, replicator._repl_to_node.call_count)
# called on +1 more replica and self when *first* handoff
replicator._repl_to_node = mock.Mock(side_effect=[DriveNotMounted()])
replicator._replicate_object('0', '/path/to/file',
replicator.ring.devs[3]['id'])
self.assertEqual(4, replicator._repl_to_node.call_count)
# even if it's the last handoff it works to keep 3 replicas
# 2 primaries + 1 handoff
replicator._repl_to_node = mock.Mock(side_effect=[DriveNotMounted()])
replicator._replicate_object('0', '/path/to/file',
replicator.ring.devs[-1]['id'])
self.assertEqual(4, replicator._repl_to_node.call_count)
# with two DriveNotMounted exceptions called on +2 more replica keeping
# durability
replicator._repl_to_node = mock.Mock(
side_effect=[DriveNotMounted()] * 2)
replicator._replicate_object('0', '/path/to/file',
replicator.ring.devs[0]['id'])
self.assertEqual(4, replicator._repl_to_node.call_count)
def test_replicate_object_with_exception_run_out_of_nodes(self):
replicator = TestReplicator({})
replicator.ring = FakeRingWithNodes().Ring('path')
replicator.brokerclass = FakeAccountBroker
replicator.delete_db = self.stub_delete_db
# all other devices are not mounted
replicator._repl_to_node = mock.Mock(side_effect=DriveNotMounted())
replicator._replicate_object('0', '/path/to/file',
replicator.ring.devs[0]['id'])
self.assertEqual(5, replicator._repl_to_node.call_count)
def test_replicate_account_out_of_place(self):
replicator = TestReplicator({}, logger=unit.FakeLogger())
replicator.ring = FakeRingWithNodes().Ring('path')
replicator.brokerclass = FakeAccountBroker
replicator._repl_to_node = lambda *args: True
replicator.delete_db = self.stub_delete_db
# Correct node_id, wrong part
part = replicator.ring.get_part(TEST_ACCOUNT_NAME) + 1
node_id = replicator.ring.get_part_nodes(part)[0]['id']
replicator._replicate_object(str(part), '/path/to/file', node_id)
self.assertEqual(['/path/to/file'], self.delete_db_calls)
error_msgs = replicator.logger.get_lines_for_level('error')
expected = 'Found /path/to/file for /a%20c%20t when it should be ' \
'on partition 0; will replicate out and remove.'
self.assertEqual(error_msgs, [expected])
def test_replicate_container_out_of_place(self):
replicator = TestReplicator({}, logger=unit.FakeLogger())
replicator.ring = FakeRingWithNodes().Ring('path')
replicator._repl_to_node = lambda *args: True
replicator.delete_db = self.stub_delete_db
# Correct node_id, wrong part
part = replicator.ring.get_part(
TEST_ACCOUNT_NAME, TEST_CONTAINER_NAME) + 1
node_id = replicator.ring.get_part_nodes(part)[0]['id']
replicator._replicate_object(str(part), '/path/to/file', node_id)
self.assertEqual(['/path/to/file'], self.delete_db_calls)
self.assertEqual(
replicator.logger.log_dict['error'],
[(('Found /path/to/file for /a%20c%20t/c%20o%20n when it should '
'be on partition 0; will replicate out and remove.',), {})])
def test_replicate_container_out_of_place_no_node(self):
replicator = TestReplicator({}, logger=unit.FakeLogger())
replicator.ring = FakeRingWithSingleNode().Ring('path')
replicator._repl_to_node = lambda *args: True
replicator.delete_db = self.stub_delete_db
# Correct node_id, wrong part
part = replicator.ring.get_part(
TEST_ACCOUNT_NAME, TEST_CONTAINER_NAME) + 1
node_id = replicator.ring.get_part_nodes(part)[0]['id']
replicator._replicate_object(str(part), '/path/to/file', node_id)
self.assertEqual(['/path/to/file'], self.delete_db_calls)
self.delete_db_calls = []
# No nodes this time
replicator.ring.get_part_nodes = lambda *args: []
replicator.delete_db = self.stub_delete_db
# Correct node_id, wrong part
part = replicator.ring.get_part(
TEST_ACCOUNT_NAME, TEST_CONTAINER_NAME) + 1
replicator._replicate_object(str(part), '/path/to/file', node_id)
self.assertEqual([], self.delete_db_calls)
def test_replicate_object_different_region(self):
db_replicator.ring = FakeRingWithNodes()
replicator = TestReplicator({})
replicator._repl_to_node = mock.Mock()
# For node_id = 1, one replica in same region(1) and other is in a
# different region(2). Refer: FakeRingWithNodes
replicator._replicate_object('0', '/path/to/file', 1)
# different_region was set True and passed to _repl_to_node()
self.assertEqual(replicator._repl_to_node.call_args_list[0][0][-1],
True)
# different_region was set False and passed to _repl_to_node()
self.assertEqual(replicator._repl_to_node.call_args_list[1][0][-1],
False)
def test_delete_db(self):
db_replicator.lock_parent_directory = lock_parent_directory
replicator = TestReplicator({}, logger=unit.FakeLogger())
replicator._zero_stats()
replicator.extract_device = lambda _: 'some_device'
temp_dir = mkdtemp()
try:
temp_suf_dir = os.path.join(temp_dir, '16e')
os.mkdir(temp_suf_dir)
temp_hash_dir = os.path.join(temp_suf_dir,
'166e33924a08ede4204871468c11e16e')
os.mkdir(temp_hash_dir)
temp_file = NamedTemporaryFile(dir=temp_hash_dir, delete=False)
temp_hash_dir2 = os.path.join(temp_suf_dir,
'266e33924a08ede4204871468c11e16e')
os.mkdir(temp_hash_dir2)
temp_file2 = NamedTemporaryFile(dir=temp_hash_dir2, delete=False)
# sanity-checks
self.assertTrue(os.path.exists(temp_dir))
self.assertTrue(os.path.exists(temp_suf_dir))
self.assertTrue(os.path.exists(temp_hash_dir))
self.assertTrue(os.path.exists(temp_file.name))
self.assertTrue(os.path.exists(temp_hash_dir2))
self.assertTrue(os.path.exists(temp_file2.name))
self.assertEqual(0, replicator.stats['remove'])
temp_file.db_file = temp_file.name
replicator.delete_db(temp_file)
self.assertTrue(os.path.exists(temp_dir))
self.assertTrue(os.path.exists(temp_suf_dir))
self.assertFalse(os.path.exists(temp_hash_dir))
self.assertFalse(os.path.exists(temp_file.name))
self.assertTrue(os.path.exists(temp_hash_dir2))
self.assertTrue(os.path.exists(temp_file2.name))
self.assertEqual([(('removes.some_device',), {})],
replicator.logger.log_dict['increment'])
self.assertEqual(1, replicator.stats['remove'])
temp_file2.db_file = temp_file2.name
replicator.delete_db(temp_file2)
self.assertTrue(os.path.exists(temp_dir))
self.assertFalse(os.path.exists(temp_suf_dir))
self.assertFalse(os.path.exists(temp_hash_dir))
self.assertFalse(os.path.exists(temp_file.name))
self.assertFalse(os.path.exists(temp_hash_dir2))
self.assertFalse(os.path.exists(temp_file2.name))
self.assertEqual([(('removes.some_device',), {})] * 2,
replicator.logger.log_dict['increment'])
self.assertEqual(2, replicator.stats['remove'])
finally:
rmtree(temp_dir)
def test_extract_device(self):
replicator = TestReplicator({'devices': '/some/root'})
self.assertEqual('some_device', replicator.extract_device(
'/some/root/some_device/deeper/and/deeper'))
self.assertEqual('UNKNOWN', replicator.extract_device(
'/some/foo/some_device/deeper/and/deeper'))
def test_dispatch_no_arg_pop(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
with unit.mock_check_drive(isdir=True):
response = rpc.dispatch(('a',), 'arg')
self.assertEqual('Invalid object type', response.body)
self.assertEqual(400, response.status_int)
def test_dispatch_drive_not_mounted(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=True)
with unit.mock_check_drive() as mocks:
response = rpc.dispatch(('drive', 'part', 'hash'), ['method'])
self.assertEqual([mock.call(os.path.join('/drive'))],
mocks['ismount'].call_args_list)
self.assertEqual('507 drive is not mounted', response.status)
self.assertEqual(507, response.status_int)
def test_dispatch_unexpected_operation_db_does_not_exist(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
def mock_mkdirs(path):
self.assertEqual('/drive/tmp', path)
self._patch(patch.object, db_replicator, 'mkdirs', mock_mkdirs)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = False
response = rpc.dispatch(('drive', 'part', 'hash'), ['unexpected'])
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
def test_dispatch_operation_unexpected(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
self._patch(patch.object, db_replicator, 'mkdirs', lambda *args: True)
def unexpected_method(broker, args):
self.assertEqual(FakeBroker, broker.__class__)
self.assertEqual(['arg1', 'arg2'], args)
return 'unexpected-called'
rpc.unexpected = unexpected_method
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = True
response = rpc.dispatch(('drive', 'part', 'hash'),
['unexpected', 'arg1', 'arg2'])
mock_os.path.exists.assert_called_with('/part/ash/hash/hash.db')
self.assertEqual('unexpected-called', response)
def test_dispatch_operation_rsync_then_merge(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
self._patch(patch.object, db_replicator, 'renamer', lambda *args: True)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = True
response = rpc.dispatch(('drive', 'part', 'hash'),
['rsync_then_merge', 'arg1', 'arg2'])
expected_calls = [call('/part/ash/hash/hash.db'),
call('/drive/tmp/arg1'),
call(FakeBroker.db_file),
call('/drive/tmp/arg1')]
self.assertEqual(mock_os.path.exists.call_args_list,
expected_calls)
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
def test_dispatch_operation_complete_rsync(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
self._patch(patch.object, db_replicator, 'renamer', lambda *args: True)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.side_effect = [False, True]
response = rpc.dispatch(('drive', 'part', 'hash'),
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
['complete_rsync', 'arg1'])
expected_calls = [call('/part/ash/hash/hash.db'),
call('/drive/tmp/arg1')]
self.assertEqual(mock_os.path.exists.call_args_list,
expected_calls)
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.side_effect = [False, True]
response = rpc.dispatch(('drive', 'part', 'hash'),
['complete_rsync', 'arg1', 'arg2'])
expected_calls = [call('/part/ash/hash/arg2'),
call('/drive/tmp/arg1')]
self.assertEqual(mock_os.path.exists.call_args_list,
expected_calls)
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
def test_rsync_then_merge_db_does_not_exist(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = False
response = rpc.rsync_then_merge('drive', '/data/db.db',
('arg1', 'arg2'))
mock_os.path.exists.assert_called_with('/data/db.db')
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
def test_rsync_then_merge_old_does_not_exist(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.side_effect = [True, False]
response = rpc.rsync_then_merge('drive', '/data/db.db',
('arg1', 'arg2'))
expected_calls = [call('/data/db.db'), call('/drive/tmp/arg1')]
self.assertEqual(mock_os.path.exists.call_args_list,
expected_calls)
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
def test_rsync_then_merge_with_objects(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
def mock_renamer(old, new):
self.assertEqual('/drive/tmp/arg1', old)
# FakeBroker uses module filename as db_file!
self.assertEqual(__file__, new)
self._patch(patch.object, db_replicator, 'renamer', mock_renamer)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = True
response = rpc.rsync_then_merge('drive', '/data/db.db',
['arg1', 'arg2'])
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
def test_complete_rsync_db_exists(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = True
response = rpc.complete_rsync('drive', '/data/db.db', ['arg1'])
mock_os.path.exists.assert_called_with('/data/db.db')
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = True
response = rpc.complete_rsync('drive', '/data/db.db',
['arg1', 'arg2'])
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
mock_os.path.exists.assert_called_with('/data/arg2')
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
def test_complete_rsync_old_file_does_not_exist(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = False
response = rpc.complete_rsync('drive', '/data/db.db',
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
['arg1'])
expected_calls = [call('/data/db.db'), call('/drive/tmp/arg1')]
self.assertEqual(expected_calls,
mock_os.path.exists.call_args_list)
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.return_value = False
response = rpc.complete_rsync('drive', '/data/db.db',
['arg1', 'arg2'])
expected_calls = [call('/data/arg2'), call('/drive/tmp/arg1')]
self.assertEqual(expected_calls,
mock_os.path.exists.call_args_list)
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
def test_complete_rsync_rename(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
def mock_renamer(old, new):
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
renamer_calls.append((old, new))
self._patch(patch.object, db_replicator, 'renamer', mock_renamer)
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
renamer_calls = []
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.side_effect = [False, True]
response = rpc.complete_rsync('drive', '/data/db.db',
['arg1'])
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
self.assertEqual(('/drive/tmp/arg1', '/data/db.db'), renamer_calls[0])
self.assertFalse(renamer_calls[1:])
renamer_calls = []
with patch('swift.common.db_replicator.os',
new=mock.MagicMock(wraps=os)) as mock_os, \
unit.mock_check_drive(isdir=True):
mock_os.path.exists.side_effect = [False, True]
response = rpc.complete_rsync('drive', '/data/db.db',
['arg1', 'arg2'])
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
self.assertEqual(('/drive/tmp/arg1', '/data/arg2'), renamer_calls[0])
self.assertFalse(renamer_calls[1:])
def test_replicator_sync_with_broker_replication_missing_table(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
rpc.logger = unit.debug_logger()
broker = FakeBroker()
broker.get_repl_missing_table = True
called = []
def mock_quarantine_db(object_file, server_type):
called.append(True)
self.assertEqual(broker.db_file, object_file)
self.assertEqual(broker.db_type, server_type)
self._patch(patch.object, db_replicator, 'quarantine_db',
mock_quarantine_db)
with unit.mock_check_drive(isdir=True):
response = rpc.sync(broker, ('remote_sync', 'hash_', 'id_',
'created_at', 'put_timestamp',
'delete_timestamp', 'metadata'))
self.assertEqual('404 Not Found', response.status)
self.assertEqual(404, response.status_int)
self.assertEqual(called, [True])
errors = rpc.logger.get_lines_for_level('error')
self.assertEqual(errors,
["Unable to decode remote metadata 'metadata'",
"Quarantining DB %s" % broker])
def test_replicator_sync(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
broker = FakeBroker()
with unit.mock_check_drive(isdir=True):
response = rpc.sync(broker, (
broker.get_sync() + 1, 12345, 'id_',
'created_at', 'put_timestamp', 'delete_timestamp',
'{"meta1": "data1", "meta2": "data2"}'))
self.assertEqual({'meta1': 'data1', 'meta2': 'data2'},
broker.metadata)
self.assertEqual('created_at', broker.created_at)
self.assertEqual('put_timestamp', broker.put_timestamp)
self.assertEqual('delete_timestamp', broker.delete_timestamp)
self.assertEqual('200 OK', response.status)
self.assertEqual(200, response.status_int)
2010-07-12 17:03:45 -05:00
def test_rsync_then_merge(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
with unit.mock_check_drive(isdir=True):
rpc.rsync_then_merge('sda1', '/srv/swift/blah', ('a', 'b'))
2010-07-12 17:03:45 -05:00
def test_merge_items(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
2010-07-12 17:03:45 -05:00
fake_broker = FakeBroker()
args = ('a', 'b')
with unit.mock_check_drive(isdir=True):
rpc.merge_items(fake_broker, args)
self.assertEqual(fake_broker.args, args)
2010-07-12 17:03:45 -05:00
def test_merge_syncs(self):
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
2010-07-12 17:03:45 -05:00
fake_broker = FakeBroker()
args = ('a', 'b')
with unit.mock_check_drive(isdir=True):
rpc.merge_syncs(fake_broker, args)
self.assertEqual(fake_broker.args, (args[0],))
2010-07-12 17:03:45 -05:00
def test_complete_rsync_with_bad_input(self):
drive = '/some/root'
db_file = __file__
args = ['old_file']
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
with unit.mock_check_drive(isdir=True):
resp = rpc.complete_rsync(drive, db_file, args)
self.assertTrue(isinstance(resp, HTTPException))
self.assertEqual(404, resp.status_int)
with unit.mock_check_drive(isdir=True):
resp = rpc.complete_rsync(drive, 'new_db_file', args)
self.assertTrue(isinstance(resp, HTTPException))
self.assertEqual(404, resp.status_int)
def test_complete_rsync(self):
drive = mkdtemp()
args = ['old_file']
rpc = db_replicator.ReplicatorRpc('/', '/', FakeBroker,
mount_check=False)
os.mkdir('%s/tmp' % drive)
old_file = '%s/tmp/old_file' % drive
new_file = '%s/new_db_file' % drive
try:
fp = open(old_file, 'w')
fp.write('void')
fp.close
resp = rpc.complete_rsync(drive, new_file, args)
self.assertEqual(204, resp.status_int)
finally:
rmtree(drive)
@unit.with_tempdir
def test_empty_suffix_and_hash_dirs_get_cleanedup(self, tempdir):
datadir = os.path.join(tempdir, 'containers')
db_path = ('450/afd/7089ab48d955ab0851fc51cc17a34afd/'
'7089ab48d955ab0851fc51cc17a34afd.db')
random_file = ('1060/xyz/1234ab48d955ab0851fc51cc17a34xyz/'
'1234ab48d955ab0851fc51cc17a34xyz.abc')
# trailing "/" indicates empty dir
paths = [
# empty part dir
'240/',
# empty suffix dir
'18/aba/',
# empty hashdir
'1054/27e/d41d8cd98f00b204e9800998ecf8427e/',
# database
db_path,
# non database file
random_file,
]
for path in paths:
path = os.path.join(datadir, path)
os.makedirs(os.path.dirname(path))
if os.path.basename(path):
# our setup requires "directories" to end in "/" (i.e. basename
# is ''); otherwise, create an empty file
open(path, 'w')
# sanity
self.assertEqual({'240', '18', '1054', '1060', '450'},
set(os.listdir(datadir)))
for path in paths:
dirpath = os.path.join(datadir, os.path.dirname(path))
self.assertTrue(os.path.isdir(dirpath))
node_id = 1
results = list(db_replicator.roundrobin_datadirs(
[(datadir, node_id, lambda p: True)]))
expected = [
('450', os.path.join(datadir, db_path), node_id),
]
self.assertEqual(results, expected)
# all the empty leaf dirs are cleaned up
for path in paths:
if os.path.basename(path):
check = self.assertTrue
else:
check = self.assertFalse
dirpath = os.path.join(datadir, os.path.dirname(path))
isdir = os.path.isdir(dirpath)
check(isdir, '%r is%s a directory!' % (
dirpath, '' if isdir else ' not'))
# despite the leaves cleaned up it takes a few loops to finish it off
self.assertEqual({'18', '1054', '1060', '450'},
set(os.listdir(datadir)))
results = list(db_replicator.roundrobin_datadirs(
[(datadir, node_id, lambda p: True)]))
self.assertEqual(results, expected)
self.assertEqual({'1054', '1060', '450'},
set(os.listdir(datadir)))
results = list(db_replicator.roundrobin_datadirs(
[(datadir, node_id, lambda p: True)]))
self.assertEqual(results, expected)
# non db file in '1060' dir is not deleted and exception is handled
self.assertEqual({'1060', '450'},
set(os.listdir(datadir)))
def test_roundrobin_datadirs(self):
listdir_calls = []
isdir_calls = []
exists_calls = []
shuffle_calls = []
rmdir_calls = []
def _listdir(path):
listdir_calls.append(path)
if not path.startswith('/srv/node/sda/containers') and \
not path.startswith('/srv/node/sdb/containers'):
return []
path = path[len('/srv/node/sdx/containers'):]
if path == '':
return ['123', '456', '789', '9999', "-5", "not-a-partition"]
# 456 will pretend to be a file
# 9999 will be an empty partition with no contents
# -5 and not-a-partition were created by something outside
# Swift
elif path == '/123':
return ['abc', 'def.db'] # def.db will pretend to be a file
elif path == '/123/abc':
# 11111111111111111111111111111abc will pretend to be a file
return ['00000000000000000000000000000abc',
'11111111111111111111111111111abc']
elif path == '/123/abc/00000000000000000000000000000abc':
return ['00000000000000000000000000000abc.db',
# This other.db isn't in the right place, so should be
# ignored later.
'000000000000000000000000000other.db',
'weird1'] # weird1 will pretend to be a dir, if asked
elif path == '/789':
return ['ghi', 'jkl'] # jkl will pretend to be a file
elif path == '/789/ghi':
# 33333333333333333333333333333ghi will pretend to be a file
return ['22222222222222222222222222222ghi',
'33333333333333333333333333333ghi']
elif path == '/789/ghi/22222222222222222222222222222ghi':
return ['22222222222222222222222222222ghi.db',
'weird2'] # weird2 will pretend to be a dir, if asked
elif path == '9999':
return []
elif path == 'not-a-partition':
raise Exception("shouldn't look in not-a-partition")
elif path == '-5':
raise Exception("shouldn't look in -5")
return []
def _isdir(path):
isdir_calls.append(path)
if not path.startswith('/srv/node/sda/containers') and \
not path.startswith('/srv/node/sdb/containers'):
return False
path = path[len('/srv/node/sdx/containers'):]
if path in ('/123', '/123/abc',
'/123/abc/00000000000000000000000000000abc',
'/123/abc/00000000000000000000000000000abc/weird1',
'/789', '/789/ghi',
'/789/ghi/22222222222222222222222222222ghi',
'/789/ghi/22222222222222222222222222222ghi/weird2',
'/9999'):
return True
return False
def _exists(arg):
exists_calls.append(arg)
return True
def _shuffle(arg):
shuffle_calls.append(arg)
def _rmdir(arg):
rmdir_calls.append(arg)
base = 'swift.common.db_replicator.'
with mock.patch(base + 'os.listdir', _listdir), \
mock.patch(base + 'os.path.isdir', _isdir), \
mock.patch(base + 'os.path.exists', _exists), \
mock.patch(base + 'random.shuffle', _shuffle), \
mock.patch(base + 'os.rmdir', _rmdir):
datadirs = [('/srv/node/sda/containers', 1, lambda p: True),
('/srv/node/sdb/containers', 2, lambda p: True)]
results = list(db_replicator.roundrobin_datadirs(datadirs))
# The results show that the .db files are returned, the devices
# interleaved.
self.assertEqual(results, [
('123', '/srv/node/sda/containers/123/abc/'
'00000000000000000000000000000abc/'
'00000000000000000000000000000abc.db', 1),
('123', '/srv/node/sdb/containers/123/abc/'
'00000000000000000000000000000abc/'
'00000000000000000000000000000abc.db', 2),
('789', '/srv/node/sda/containers/789/ghi/'
'22222222222222222222222222222ghi/'
'22222222222222222222222222222ghi.db', 1),
('789', '/srv/node/sdb/containers/789/ghi/'
'22222222222222222222222222222ghi/'
'22222222222222222222222222222ghi.db', 2)])
# The listdir calls show that we only listdir the dirs
self.assertEqual(listdir_calls, [
'/srv/node/sda/containers',
'/srv/node/sda/containers/123',
'/srv/node/sda/containers/123/abc',
'/srv/node/sdb/containers',
'/srv/node/sdb/containers/123',
'/srv/node/sdb/containers/123/abc',
'/srv/node/sda/containers/789',
'/srv/node/sda/containers/789/ghi',
'/srv/node/sdb/containers/789',
'/srv/node/sdb/containers/789/ghi',
'/srv/node/sda/containers/9999',
'/srv/node/sdb/containers/9999'])
# The isdir calls show that we did ask about the things pretending
# to be files at various levels.
self.assertEqual(isdir_calls, [
'/srv/node/sda/containers/123',
'/srv/node/sda/containers/123/abc',
('/srv/node/sda/containers/123/abc/'
'00000000000000000000000000000abc'),
'/srv/node/sdb/containers/123',
'/srv/node/sdb/containers/123/abc',
('/srv/node/sdb/containers/123/abc/'
'00000000000000000000000000000abc'),
('/srv/node/sda/containers/123/abc/'
'11111111111111111111111111111abc'),
'/srv/node/sda/containers/123/def.db',
'/srv/node/sda/containers/456',
'/srv/node/sda/containers/789',
'/srv/node/sda/containers/789/ghi',
('/srv/node/sda/containers/789/ghi/'
'22222222222222222222222222222ghi'),
('/srv/node/sdb/containers/123/abc/'
'11111111111111111111111111111abc'),
'/srv/node/sdb/containers/123/def.db',
'/srv/node/sdb/containers/456',
'/srv/node/sdb/containers/789',
'/srv/node/sdb/containers/789/ghi',
('/srv/node/sdb/containers/789/ghi/'
'22222222222222222222222222222ghi'),
('/srv/node/sda/containers/789/ghi/'
'33333333333333333333333333333ghi'),
'/srv/node/sda/containers/789/jkl',
'/srv/node/sda/containers/9999',
('/srv/node/sdb/containers/789/ghi/'
'33333333333333333333333333333ghi'),
'/srv/node/sdb/containers/789/jkl',
'/srv/node/sdb/containers/9999'])
# The exists calls are the .db files we looked for as we walked the
# structure.
self.assertEqual(exists_calls, [
('/srv/node/sda/containers/123/abc/'
'00000000000000000000000000000abc/'
'00000000000000000000000000000abc.db'),
('/srv/node/sdb/containers/123/abc/'
'00000000000000000000000000000abc/'
'00000000000000000000000000000abc.db'),
('/srv/node/sda/containers/789/ghi/'
'22222222222222222222222222222ghi/'
'22222222222222222222222222222ghi.db'),
('/srv/node/sdb/containers/789/ghi/'
'22222222222222222222222222222ghi/'
'22222222222222222222222222222ghi.db')])
# Shows that we called shuffle twice, once for each device.
self.assertEqual(
shuffle_calls, [['123', '456', '789', '9999'],
['123', '456', '789', '9999']])
# Shows that we called removed the two empty partition directories.
self.assertEqual(
rmdir_calls, ['/srv/node/sda/containers/9999',
'/srv/node/sdb/containers/9999'])
@mock.patch("swift.common.db_replicator.ReplConnection", mock.Mock())
def test_http_connect(self):
node = "node"
partition = "partition"
db_file = __file__
replicator = TestReplicator({})
replicator._http_connect(node, partition, db_file)
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
expected_hsh = os.path.basename(db_file).split('.', 1)[0]
expected_hsh = expected_hsh.split('_', 1)[0]
db_replicator.ReplConnection.assert_has_calls([
Add support for sharding in ContainerBroker With this patch the ContainerBroker gains several new features: 1. A shard_ranges table to persist ShardRange data, along with methods to merge and access ShardRange instances to that table, and to remove expired shard ranges. 2. The ability to create a fresh db file to replace the existing db file. Fresh db files are named using the hash of the container path plus an epoch which is a serialized Timestamp value, in the form: <hash>_<epoch>.db During sharding both the fresh and retiring db files co-exist on disk. The ContainerBroker is now able to choose the newest on disk db file when instantiated. It also provides a method (get_brokers()) to gain access to broker instance for either on disk file. 3. Methods to access the current state of the on disk db files i.e. UNSHARDED (old file only), SHARDING (fresh and retiring files), or SHARDED (fresh file only with shard ranges). Container replication is also modified: 1. shard ranges are replicated between container db peers. Unlike objects, shard ranges are both pushed and pulled during a REPLICATE event. 2. If a container db is capable of being sharded (i.e. it has a set of shard ranges) then it will no longer attempt to replicate objects to its peers. Object record durability is achieved by sharding rather than peer to peer replication. Co-Authored-By: Matthew Oliver <matt@oliver.net.au> Co-Authored-By: Tim Burke <tim.burke@gmail.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Ie4d2816259e6c25c346976e181fb9d350f947190
2018-05-01 15:44:18 +01:00
mock.call(node, partition, expected_hsh, replicator.logger)])
class TestHandoffsOnly(unittest.TestCase):
class FakeRing3Nodes(object):
_replicas = 3
# Three nodes, two disks each
devs = [
dict(id=0, region=1, zone=1,
meta='', weight=500.0, ip='10.0.0.1', port=6201,
replication_ip='10.0.0.1', replication_port=6201,
device='sdp'),
dict(id=1, region=1, zone=1,
meta='', weight=500.0, ip='10.0.0.1', port=6201,
replication_ip='10.0.0.1', replication_port=6201,
device='sdq'),
dict(id=2, region=1, zone=1,
meta='', weight=500.0, ip='10.0.0.2', port=6201,
replication_ip='10.0.0.2', replication_port=6201,
device='sdp'),
dict(id=3, region=1, zone=1,
meta='', weight=500.0, ip='10.0.0.2', port=6201,
replication_ip='10.0.0.2', replication_port=6201,
device='sdq'),
dict(id=4, region=1, zone=1,
meta='', weight=500.0, ip='10.0.0.3', port=6201,
replication_ip='10.0.0.3', replication_port=6201,
device='sdp'),
dict(id=5, region=1, zone=1,
meta='', weight=500.0, ip='10.0.0.3', port=6201,
replication_ip='10.0.0.3', replication_port=6201,
device='sdq'),
]
def __init__(self, *a, **kw):
pass
def get_part(self, account, container=None, obj=None):
return 0
def get_part_nodes(self, part):
nodes = []
for offset in range(self._replicas):
i = (part + offset) % len(self.devs)
nodes.append(self.devs[i])
return nodes
def get_more_nodes(self, part):
for offset in range(self._replicas, len(self.devs)):
i = (part + offset) % len(self.devs)
yield self.devs[i]
def _make_fake_db(self, disk, partition, db_hash):
directories = [
os.path.join(self.root, disk),
os.path.join(self.root, disk, 'containers'),
os.path.join(self.root, disk, 'containers', str(partition)),
os.path.join(self.root, disk, 'containers', str(partition),
db_hash[-3:]),
os.path.join(self.root, disk, 'containers', str(partition),
db_hash[-3:], db_hash)]
for d in directories:
try:
os.mkdir(d)
except OSError as err:
if err.errno != errno.EEXIST:
raise
file_path = os.path.join(directories[-1], db_hash + ".db")
with open(file_path, 'w'):
pass
def setUp(self):
self.root = mkdtemp()
# object disks; they're just here to make sure they don't trip us up
os.mkdir(os.path.join(self.root, 'sdc'))
os.mkdir(os.path.join(self.root, 'sdc', 'objects'))
os.mkdir(os.path.join(self.root, 'sdd'))
os.mkdir(os.path.join(self.root, 'sdd', 'objects'))
# part 0 belongs on sdp
self._make_fake_db('sdp', 0, '010101013cf2b7979af9eaa71cb67220')
# part 1 does not belong on sdp
self._make_fake_db('sdp', 1, 'abababab2b5368158355e799323b498d')
# part 1 belongs on sdq
self._make_fake_db('sdq', 1, '02020202e30f696a3cfa63d434a3c94e')
# part 2 does not belong on sdq
self._make_fake_db('sdq', 2, 'bcbcbcbc15d3835053d568c57e2c83b5')
def cleanUp(self):
rmtree(self.root, ignore_errors=True)
def test_scary_warnings(self):
logger = unit.FakeLogger()
replicator = TestReplicator({
'handoffs_only': 'yes',
'devices': self.root,
'bind_port': 6201,
'mount_check': 'no',
}, logger=logger)
with patch.object(db_replicator, 'whataremyips',
return_value=['10.0.0.1']), \
patch.object(replicator, '_replicate_object'), \
patch.object(replicator, 'ring', self.FakeRing3Nodes()):
replicator.run_once()
self.assertEqual(
logger.get_lines_for_level('warning'),
[('Starting replication pass with handoffs_only enabled. This '
'mode is not intended for normal operation; use '
'handoffs_only with care.'),
('Finished replication pass with handoffs_only enabled. '
'If handoffs_only is no longer required, disable it.')])
def test_skips_primary_partitions(self):
replicator = TestReplicator({
'handoffs_only': 'yes',
'devices': self.root,
'bind_port': 6201,
'mount_check': 'no',
})
with patch.object(db_replicator, 'whataremyips',
return_value=['10.0.0.1']), \
patch.object(replicator, '_replicate_object') as mock_repl, \
patch.object(replicator, 'ring', self.FakeRing3Nodes()):
replicator.run_once()
self.assertEqual(sorted(mock_repl.mock_calls), [
mock.call('1', os.path.join(
self.root, 'sdp', 'containers', '1', '98d',
'abababab2b5368158355e799323b498d',
'abababab2b5368158355e799323b498d.db'), 0),
mock.call('2', os.path.join(
self.root, 'sdq', 'containers', '2', '3b5',
'bcbcbcbc15d3835053d568c57e2c83b5',
'bcbcbcbc15d3835053d568c57e2c83b5.db'), 1)])
def test_override_partitions(self):
replicator = TestReplicator({
'devices': self.root,
'bind_port': 6201,
'mount_check': 'no',
})
with patch.object(db_replicator, 'whataremyips',
return_value=['10.0.0.1']), \
patch.object(replicator, '_replicate_object') as mock_repl, \
patch.object(replicator, 'ring', self.FakeRing3Nodes()):
replicator.run_once(partitions="0,2")
self.assertEqual(sorted(mock_repl.mock_calls), [
mock.call('0', os.path.join(
self.root, 'sdp', 'containers', '0', '220',
'010101013cf2b7979af9eaa71cb67220',
'010101013cf2b7979af9eaa71cb67220.db'), 0),
mock.call('2', os.path.join(
self.root, 'sdq', 'containers', '2', '3b5',
'bcbcbcbc15d3835053d568c57e2c83b5',
'bcbcbcbc15d3835053d568c57e2c83b5.db'), 1)])
def test_override_devices(self):
replicator = TestReplicator({
'devices': self.root,
'bind_port': 6201,
'mount_check': 'no',
})
with patch.object(db_replicator, 'whataremyips',
return_value=['10.0.0.1']), \
patch.object(replicator, '_replicate_object') as mock_repl, \
patch.object(replicator, 'ring', self.FakeRing3Nodes()):
replicator.run_once(devices="sdp")
self.assertEqual(sorted(mock_repl.mock_calls), [
mock.call('0', os.path.join(
self.root, 'sdp', 'containers', '0', '220',
'010101013cf2b7979af9eaa71cb67220',
'010101013cf2b7979af9eaa71cb67220.db'), 0),
mock.call('1', os.path.join(
self.root, 'sdp', 'containers', '1', '98d',
'abababab2b5368158355e799323b498d',
'abababab2b5368158355e799323b498d.db'), 0)])
def test_override_devices_and_partitions(self):
replicator = TestReplicator({
'devices': self.root,
'bind_port': 6201,
'mount_check': 'no',
})
with patch.object(db_replicator, 'whataremyips',
return_value=['10.0.0.1']), \
patch.object(replicator, '_replicate_object') as mock_repl, \
patch.object(replicator, 'ring', self.FakeRing3Nodes()):
replicator.run_once(partitions="0,2", devices="sdp")
self.assertEqual(sorted(mock_repl.mock_calls), [
mock.call('0', os.path.join(
self.root, 'sdp', 'containers', '0', '220',
'010101013cf2b7979af9eaa71cb67220',
'010101013cf2b7979af9eaa71cb67220.db'), 0)])
class TestReplToNode(unittest.TestCase):
def setUp(self):
db_replicator.ring = FakeRing()
self.delete_db_calls = []
self.broker = FakeBroker()
self.replicator = TestReplicator({'per_diff': 10})
self.fake_node = {'ip': '127.0.0.1', 'device': 'sda1', 'port': 1000}
self.fake_info = {'id': 'a', 'point': -1, 'max_row': 20, 'hash': 'b',
'created_at': 100, 'put_timestamp': 0,
'delete_timestamp': 0, 'count': 0,
'metadata': json.dumps({
'Test': ('Value', normalize_timestamp(1))})}
self.replicator.logger = mock.Mock()
self.replicator._rsync_db = mock.Mock(return_value=True)
self.replicator._usync_db = mock.Mock(return_value=True)
self.http = ReplHttp('{"id": 3, "point": -1}')
self.replicator._http_connect = lambda *args: self.http
def test_repl_to_node_usync_success(self):
rinfo = {"id": 3, "point": -1, "max_row": 10, "hash": "c"}
self.http = ReplHttp(json.dumps(rinfo))
local_sync = self.broker.get_sync()
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, self.broker, '0', self.fake_info), True)
self.replicator._usync_db.assert_has_calls([
mock.call(max(rinfo['point'], local_sync), self.broker,
self.http, rinfo['id'], self.fake_info['id'])
])
def test_repl_to_node_rsync_success(self):
rinfo = {"id": 3, "point": -1, "max_row": 9, "hash": "c"}
self.http = ReplHttp(json.dumps(rinfo))
self.broker.get_sync()
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, self.broker, '0', self.fake_info), True)
self.replicator.logger.increment.assert_has_calls([
mock.call.increment('remote_merges')
])
self.replicator._rsync_db.assert_has_calls([
mock.call(self.broker, self.fake_node, self.http,
self.fake_info['id'],
replicate_method='rsync_then_merge',
replicate_timeout=(self.fake_info['count'] / 2000),
different_region=False)
])
def test_repl_to_node_already_in_sync(self):
rinfo = {"id": 3, "point": -1, "max_row": 20, "hash": "b"}
self.http = ReplHttp(json.dumps(rinfo))
self.broker.get_sync()
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, self.broker, '0', self.fake_info), True)
self.assertEqual(self.replicator._rsync_db.call_count, 0)
self.assertEqual(self.replicator._usync_db.call_count, 0)
def test_repl_to_node_metadata_update(self):
now = Timestamp(time.time()).internal
rmetadata = {"X-Container-Sysmeta-Test": ("XYZ", now)}
rinfo = {"id": 3, "point": -1, "max_row": 20, "hash": "b",
"metadata": json.dumps(rmetadata)}
self.http = ReplHttp(json.dumps(rinfo))
self.broker.get_sync()
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, self.broker, '0', self.fake_info), True)
metadata = self.broker.metadata
self.assertIn("X-Container-Sysmeta-Test", metadata)
self.assertEqual("XYZ", metadata["X-Container-Sysmeta-Test"][0])
self.assertEqual(now, metadata["X-Container-Sysmeta-Test"][1])
def test_repl_to_node_not_found(self):
self.http = ReplHttp('{"id": 3, "point": -1}', set_status=404)
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, self.broker, '0', self.fake_info, False), True)
self.replicator.logger.increment.assert_has_calls([
mock.call.increment('rsyncs')
])
self.replicator._rsync_db.assert_has_calls([
mock.call(self.broker, self.fake_node, self.http,
self.fake_info['id'], different_region=False)
])
def test_repl_to_node_drive_not_mounted(self):
self.http = ReplHttp('{"id": 3, "point": -1}', set_status=507)
self.assertRaises(DriveNotMounted, self.replicator._repl_to_node,
self.fake_node, FakeBroker(), '0', self.fake_info)
def test_repl_to_node_300_status(self):
self.http = ReplHttp('{"id": 3, "point": -1}', set_status=300)
self.assertFalse(self.replicator._repl_to_node(
self.fake_node, FakeBroker(), '0', self.fake_info))
def test_repl_to_node_not_response(self):
self.http = mock.Mock(replicate=mock.Mock(return_value=None))
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, FakeBroker(), '0', self.fake_info), False)
def test_repl_to_node_small_container_always_usync(self):
# Tests that a small container that is > 50% out of sync will
# still use usync.
rinfo = {"id": 3, "point": -1, "hash": "c"}
# Turn per_diff back to swift's default.
self.replicator.per_diff = 1000
for r, l in ((5, 20), (40, 100), (450, 1000), (550, 1500)):
rinfo['max_row'] = r
self.fake_info['max_row'] = l
self.replicator._usync_db = mock.Mock(return_value=True)
self.http = ReplHttp(json.dumps(rinfo))
local_sync = self.broker.get_sync()
self.assertEqual(self.replicator._repl_to_node(
self.fake_node, self.broker, '0', self.fake_info), True)
self.replicator._usync_db.assert_has_calls([
mock.call(max(rinfo['point'], local_sync), self.broker,
self.http, rinfo['id'], self.fake_info['id'])
])
class ExampleReplicator(db_replicator.Replicator):
server_type = 'fake'
brokerclass = ExampleBroker
datadir = 'fake'
default_port = 1000
class TestReplicatorSync(unittest.TestCase):
# override in subclass
backend = ExampleReplicator.brokerclass
datadir = ExampleReplicator.datadir
replicator_daemon = ExampleReplicator
replicator_rpc = db_replicator.ReplicatorRpc
def setUp(self):
self.root = mkdtemp()
self.rpc = self.replicator_rpc(
self.root, self.datadir, self.backend, mount_check=False,
logger=unit.debug_logger())
FakeReplConnection = attach_fake_replication_rpc(self.rpc)
self._orig_ReplConnection = db_replicator.ReplConnection
db_replicator.ReplConnection = FakeReplConnection
self._orig_Ring = db_replicator.ring.Ring
self._ring = unit.FakeRing()
db_replicator.ring.Ring = lambda *args, **kwargs: self._get_ring()
self.logger = unit.debug_logger()
def tearDown(self):
db_replicator.ReplConnection = self._orig_ReplConnection
db_replicator.ring.Ring = self._orig_Ring
rmtree(self.root)
def _get_ring(self):
return self._ring
def _get_broker(self, account, container=None, node_index=0):
hash_ = hash_path(account, container)
part, nodes = self._ring.get_nodes(account, container)
drive = nodes[node_index]['device']
db_path = os.path.join(self.root, drive,
storage_directory(self.datadir, part, hash_),
hash_ + '.db')
return self.backend(db_path, account=account, container=container)
def _get_broker_part_node(self, broker):
part, nodes = self._ring.get_nodes(broker.account, broker.container)
storage_dir = broker.db_file[len(self.root):].lstrip(os.path.sep)
broker_device = storage_dir.split(os.path.sep, 1)[0]
for node in nodes:
if node['device'] == broker_device:
return part, node
Fix large out of sync out of date containers As I understand it db replication starts with a preflight sync request to the remote container server who's response will include the last synced row_id that it has on file for the sending nodes database id. If the difference in the last sync point returned is more than 50% of the local sending db's rows, it'll fall back to sending the whole db over rsync and let the remote end merge items locally - but generally there's just a few rows missing and they're shipped over the wire as json and stuffed into some rather normal looking merge_items calls. The one thing that's a bit different with these remote merge_items calls (compared to your average run of the mill eat a bunch of entries out of a .pending file) is the is source kwarg. When this optional kwarg comes into merge_items it's the remote sending db's uuid, and after we eat all the rows it sent us we update our local incoming_sync table for that uuid so that next time when it makes it's pre-flight sync request we can tell it where it left off. Now normally the sending db is going to push out it's rows up from the returned sync_point in 1000 item diffs, up to 10 batches total (per_diff and max_diffs options) - 10K rows. If that goes well then everything is in sync up to at least the point it started, and the sending db will *also* ship over *it's* incoming_sync rows to merge_syncs on the remote end. Since the sending db is in sync with these other db's up to those points so is the remote db now by way of the transitive property. Also note through some weird artifact that I'm not entirely convinced isn't an unrelated and possibly benign bug the incoming_sync table on the sending db will often also happen to include it's own uuid - maybe it got pushed back to it from another node? Anyway, that seemed to work well enough until a sending db got diff capped (i.e. sent it's 10K rows and wasn't finished), when this happened the final merge_syncs call never gets sent because the remote end is definitely *not* up to date with the other databases that the sending db is - it's not even up-to-date with the sending db yet! But the hope is certainly that on the next pass it'll be able to finish sending the remaining items. But since the remote end is who decides what the last successfully synced row with this local sending db was - it's super important that the incoming_sync table is getting updated in merge_items when that source kwarg is there. I observed this simple and straight forward process wasn't working well in one case - which is weird considering it didn't have much in the way of tests. After I had the test and started looking into it seemed maybe the source kwarg handling got over-indented a bit in the bulk insert merge_items refactor. I think this is correct - maybe we could send someone up to the mountain temple to seek out gholt? Change-Id: I4137388a97925814748ecc36b3ab5f1ac3309659
2014-12-11 01:59:52 -08:00
def _get_daemon(self, node, conf_updates):
conf = {
'devices': self.root,
'recon_cache_path': self.root,
'mount_check': 'false',
'bind_port': node['replication_port'],
}
if conf_updates:
conf.update(conf_updates)
Fix large out of sync out of date containers As I understand it db replication starts with a preflight sync request to the remote container server who's response will include the last synced row_id that it has on file for the sending nodes database id. If the difference in the last sync point returned is more than 50% of the local sending db's rows, it'll fall back to sending the whole db over rsync and let the remote end merge items locally - but generally there's just a few rows missing and they're shipped over the wire as json and stuffed into some rather normal looking merge_items calls. The one thing that's a bit different with these remote merge_items calls (compared to your average run of the mill eat a bunch of entries out of a .pending file) is the is source kwarg. When this optional kwarg comes into merge_items it's the remote sending db's uuid, and after we eat all the rows it sent us we update our local incoming_sync table for that uuid so that next time when it makes it's pre-flight sync request we can tell it where it left off. Now normally the sending db is going to push out it's rows up from the returned sync_point in 1000 item diffs, up to 10 batches total (per_diff and max_diffs options) - 10K rows. If that goes well then everything is in sync up to at least the point it started, and the sending db will *also* ship over *it's* incoming_sync rows to merge_syncs on the remote end. Since the sending db is in sync with these other db's up to those points so is the remote db now by way of the transitive property. Also note through some weird artifact that I'm not entirely convinced isn't an unrelated and possibly benign bug the incoming_sync table on the sending db will often also happen to include it's own uuid - maybe it got pushed back to it from another node? Anyway, that seemed to work well enough until a sending db got diff capped (i.e. sent it's 10K rows and wasn't finished), when this happened the final merge_syncs call never gets sent because the remote end is definitely *not* up to date with the other databases that the sending db is - it's not even up-to-date with the sending db yet! But the hope is certainly that on the next pass it'll be able to finish sending the remaining items. But since the remote end is who decides what the last successfully synced row with this local sending db was - it's super important that the incoming_sync table is getting updated in merge_items when that source kwarg is there. I observed this simple and straight forward process wasn't working well in one case - which is weird considering it didn't have much in the way of tests. After I had the test and started looking into it seemed maybe the source kwarg handling got over-indented a bit in the bulk insert merge_items refactor. I think this is correct - maybe we could send someone up to the mountain temple to seek out gholt? Change-Id: I4137388a97925814748ecc36b3ab5f1ac3309659
2014-12-11 01:59:52 -08:00
return self.replicator_daemon(conf, logger=self.logger)
def _install_fake_rsync_file(self, daemon, captured_calls=None):
def _rsync_file(db_file, remote_file, **kwargs):
if captured_calls is not None:
captured_calls.append((db_file, remote_file, kwargs))
remote_server, remote_path = remote_file.split('/', 1)
dest_path = os.path.join(self.root, remote_path)
copy(db_file, dest_path)
return True
daemon._rsync_file = _rsync_file
def _run_once(self, node, conf_updates=None, daemon=None):
daemon = daemon or self._get_daemon(node, conf_updates)
self._install_fake_rsync_file(daemon)
with mock.patch('swift.common.db_replicator.whataremyips',
new=lambda *a, **kw: [node['replication_ip']]), \
unit.mock_check_drive(isdir=not daemon.mount_check,
ismount=daemon.mount_check):
daemon.run_once()
return daemon
def test_local_ids(self):
for drive in ('sda', 'sdb', 'sdd'):
os.makedirs(os.path.join(self.root, drive, self.datadir))
for node in self._ring.devs:
daemon = self._run_once(node)
if node['device'] == 'sdc':
self.assertEqual(daemon._local_device_ids, set())
else:
self.assertEqual(daemon._local_device_ids,
set([node['id']]))
def test_clean_up_after_deleted_brokers(self):
broker = self._get_broker('a', 'c', node_index=0)
part, node = self._get_broker_part_node(broker)
part = str(part)
daemon = self._run_once(node)
# create a super old broker and delete it!
forever_ago = time.time() - daemon.reclaim_age
put_timestamp = normalize_timestamp(forever_ago - 2)
delete_timestamp = normalize_timestamp(forever_ago - 1)
broker.initialize(put_timestamp)
broker.delete_db(delete_timestamp)
# if we have a container broker make sure it's reported
if hasattr(broker, 'reported'):
info = broker.get_info()
broker.reported(info['put_timestamp'],
info['delete_timestamp'],
info['object_count'],
info['bytes_used'])
info = broker.get_replication_info()
self.assertTrue(daemon.report_up_to_date(info))
# we have a part dir
part_root = os.path.join(self.root, node['device'], self.datadir)
parts = os.listdir(part_root)
self.assertEqual([part], parts)
# with a single suffix
suff = os.listdir(os.path.join(part_root, part))
self.assertEqual(1, len(suff))
# running replicator will remove the deleted db
daemon = self._run_once(node, daemon=daemon)
self.assertEqual(1, daemon.stats['remove'])
# we still have a part dir (but it's empty)
suff = os.listdir(os.path.join(part_root, part))
self.assertEqual(0, len(suff))
# run it again and there's nothing to do...
daemon = self._run_once(node, daemon=daemon)
self.assertEqual(0, daemon.stats['attempted'])
# but empty part dir is cleaned up!
parts = os.listdir(part_root)
self.assertEqual(0, len(parts))
def test_rsync_then_merge(self):
# setup current db (and broker)
broker = self._get_broker('a', 'c', node_index=0)
part, node = self._get_broker_part_node(broker)
part = str(part)
put_timestamp = normalize_timestamp(time.time())
broker.initialize(put_timestamp)
put_metadata = {'example-meta': ['bah', put_timestamp]}
broker.update_metadata(put_metadata)
# sanity (re-open, and the db keeps the metadata)
broker = self._get_broker('a', 'c', node_index=0)
self.assertEqual(put_metadata, broker.metadata)
# create rsynced db in tmp dir
obj_hash = hash_path('a', 'c')
rsynced_db_broker = self.backend(
os.path.join(self.root, node['device'], 'tmp', obj_hash + '.db'),
account='a', container='b')
rsynced_db_broker.initialize(put_timestamp)
# do rysnc_then_merge
rpc = db_replicator.ReplicatorRpc(
self.root, self.datadir, self.backend, False)
response = rpc.dispatch((node['device'], part, obj_hash),
['rsync_then_merge', obj_hash + '.db', 'arg2'])
# sanity
self.assertEqual('204 No Content', response.status)
self.assertEqual(204, response.status_int)
# re-open the db
broker = self._get_broker('a', 'c', node_index=0)
# keep the metadata in existing db
self.assertEqual(put_metadata, broker.metadata)
def test_replicator_sync(self):
# setup current db (and broker)
broker = self._get_broker('a', 'c', node_index=0)
part, node = self._get_broker_part_node(broker)
part = str(part)
put_timestamp = normalize_timestamp(time.time())
broker.initialize(put_timestamp)
put_metadata = {'example-meta': ['bah', put_timestamp]}
sync_local_metadata = {
"meta1": ["data1", put_timestamp],
"meta2": ["data2", put_timestamp]}
broker.update_metadata(put_metadata)
# sanity (re-open, and the db keeps the metadata)
broker = self._get_broker('a', 'c', node_index=0)
self.assertEqual(put_metadata, broker.metadata)
# do rysnc_then_merge
rpc = db_replicator.ReplicatorRpc(
self.root, self.datadir, ExampleBroker, False)
response = rpc.sync(
broker, (broker.get_sync('id_') + 1, 12345, 'id_',
put_timestamp, put_timestamp, '0',
json.dumps(sync_local_metadata)))
# sanity
self.assertEqual('200 OK', response.status)
self.assertEqual(200, response.status_int)
# re-open the db
broker = self._get_broker('a', 'c', node_index=0)
# keep the both metadata in existing db and local db
expected = put_metadata.copy()
expected.update(sync_local_metadata)
self.assertEqual(expected, broker.metadata)
2010-07-12 17:03:45 -05:00
if __name__ == '__main__':
unittest.main()