1、Replicator执行代码详细分析
上篇问中介绍了启动Replicator的具体过程，下面讲解Replicator的执行代码的具体实现，首先看replicate方法：
def replicate(self, override_devices=None, override_partitions=None):
"""Run a replication pass"""
self.start = time.time()
self.suffix_count = 0
self.suffix_sync = 0
self.suffix_hash = 0
self.replication_count = 0
self.last_replication_count = -1
self.partition_times = []
if override_devices is None:
override_devices = []
if override_partitions is None:
override_partitions = []
#heartbeat 为心跳函数 根据配置，配置没有 默认为 300
stats = eventlet.spawn(self.heartbeat)
#detect_lockup 检查死锁
lockup_detector = eventlet.spawn(self.detect_lockups)
eventlet.sleep() # Give spawns a cycle
try:
#replication 的 woker 数量
self.run_pool = GreenPool(size=self.concurrency)
# Returns a sorted list of jobs (dictionaries) that specify the
# partitions, nodes, etc to be synced.
# 返回专门为分区，节点同步工作的排序的列表
#
jobs = self.collect_jobs()
for job in jobs:
#重写设备
if override_devices and job['device'] not in override_devices:
continue
#重写分区
if override_partitions and \
job['partition'] not in override_partitions:
continue
#如果重写设备及其重写分区在job 中
dev_path = join(self.devices_dir, job['device'])
if self.mount_check and not ismount(dev_path):
self.logger.warn(_('%s is not mounted'), job['device'])
continue
#ring没有改变
if not self.check_ring():
self.logger.info(_("Ring change detected. Aborting "
"current replication pass."))
return
#如果
if job['delete']:
self.run_pool.spawn(self.update_deleted, job)
else:
#执行的是更新
self.run_pool.spawn(self.update, job)
with Timeout(self.lockup_timeout):
self.run_pool.waitall()
except (Exception, Timeout):
self.logger.exception(_("Exception in top-level replication loop"))
self.kill_coros()
finally:
stats.kill()
lockup_detector.kill()
self.stats_line()
在replicate方法中，首先是为replicate方法执行的准备工作，其中最重要的是要收集要执行的job的collection_jobs方法，下面为其代码的具体实现：
def collect_jobs(self):
"""
Returns a sorted list of jobs (dictionaries) that specify the
partitions, nodes, etc to be synced.
"""
jobs = []
ips = whataremyips()
#replication_ip 和replication_port 在 RingBuilder中 load添加
#self.object_ring = Ring(self.swift_dir, ring_name='object')
for local_dev in [dev for dev in self.object_ring.devs
if dev and dev['replication_ip'] in ips and
dev['replication_port'] == self.port]:
dev_path = join(self.devices_dir, local_dev['device'])
obj_path = join(dev_path, 'objects')
tmp_path = join(dev_path, 'tmp')
if self.mount_check and not ismount(dev_path):
self.logger.warn(_('%s is not mounted'), local_dev['device'])
continue
#Remove any file in a given path that that was last modified before mtime.
#/srv/1/node/sdb1/tmp下的文件
unlink_older_than(tmp_path, time.time() - self.reclaim_age)
if not os.path.exists(obj_path):
try:
mkdirs(obj_path)
except Exception:
self.logger.exception('ERROR creating %s' % obj_path)
continue
#root@kinglion-Lenovo-Product:/srv/1/node/sdb1/objects# ls
#13069 133971 4799 58208 94238
for partition in os.listdir(obj_path):
try:
job_path = join(obj_path, partition)
#判断当前路径是否为文件，如果是文件则删除
if isfile(job_path):
#
# Clean up any (probably zero-byte) files where a
# partition should be.
self.logger.warning('Removing partition directory '
'which was a file: %s', job_path)
os.remove(job_path)
continue
#获得每个partion对应的设备
part_nodes = \
self.object_ring.get_part_nodes(int(partition))
#nodes为不是本机器nodes的其他replica-1个nodes
nodes = [node for node in part_nodes
if node['id'] != local_dev['id']]
#对objects下所有partion遍历，故有jobs的长度最大为_replica2part2dev分区备份中出现此设备有此设备id的分区和
jobs.append(
dict(path=job_path,
device=local_dev['device'],
nodes=nodes,
#len（nodes）>len(part_nodes)-1的情况是当前节点已经不再是 当前partition所对应的设备了，有可能删除了该设备
delete=len(nodes) > len(part_nodes) - 1,
partition=partition))
except (ValueError, OSError):
continue
#打乱顺序
random.shuffle(jobs)
if self.handoffs_first:
# Move the handoff parts to the front of the list
#将handoff 节点移到jobs队列的前边
jobs.sort(key=lambda job: not job['delete'])
self.job_count = len(jobs)
return jobs对于第二层for循环，os.listdir(obj_path)列出objects目录下的所有partion，创建object是在objects文件夹下创建objects所映射的分区号的文件件，再在partion文件夹下创建以object的hash值后三位为名称的文件夹，然后再在后缀文件夹下创建以object的hash值为文件夹名的文件夹，object会存储为以object上传时间戳为名.data为文件后缀的文件。通过理解一致性hash算法可知，加入虚拟节点后每一个设备会多个虚拟节点和其对应，如果一个设备对应的分区为n则，obj_path下子文件夹数目会<=n,因为存入的所有文件并不一定都能映射到当前设备所对应的分区。for循环首先判读obj_path下是否为文件，若是文件则删除，若不是则获得该分区号，根据分区号获得该分区号所映射的三个备份设备，并将设备id和本地设备id不想等的加入到nodes中，将nodes、path等信息加入到jobs中，最后打乱jobs的顺序，再将handoff 节点移到队列前边。返回jobs。再到replicate方法，首先我们看job[delete]为False的情况。当job[delete]为False会执行update方法，下边看update方法的具体实现：
def update(self, job):
"""
High-level method that replicates a single partition.
:param job: a dict containing info about the partition to be replicated
"""
self.replication_count += 1
self.logger.increment('partition.update.count.%s' % (job['device'],))
begin = time.time()
try:
#get_hashes 从hashes.pkl获取hashes值并更新 获取本地的hashes job[path] 为 job_path = join(obj_path, partition) local_hash为hashes.pkl中的反序列化回来的内容 hashed为改变的
hashed, local_hash = tpool_reraise(
get_hashes, job['path'],
do_listdir=(self.replication_count % 10) == 0,
reclaim_age=self.reclaim_age)
self.suffix_hash += hashed
self.logger.update_stats('suffix.hashes', hashed)
#
attempts_left = len(job['nodes'])
#此时的nodes为除去本节点外的所有节点 因为 job['nodes]不包含本地节点get_more_nodes(int(job['partition']))能获得除去本partion所对应节点 外的其他所有节点
nodes = itertools.chain(
job['nodes'],
self.object_ring.get_more_nodes(int(job['partition'])))
#此时attempts_left 为2 若果replica为3
while attempts_left > 0:
# If this throws StopIterator it will be caught way below
node = next(nodes)
attempts_left -= 1
try:
with Timeout(self.http_timeout):
#REPLICARE方法 对应 sever里面的RELICATE方法
resp = http_connect(
node['replication_ip'], node['replication_port'],
node['device'], job['partition'], 'REPLICATE',
'', headers=self.headers).getresponse()
if resp.status == HTTP_INSUFFICIENT_STORAGE:
self.logger.error(_('%(ip)s/%(device)s responded'
' as unmounted'), node)
attempts_left += 1
continue
if resp.status != HTTP_OK:
self.logger.error(_("Invalid response %(resp)s "
"from %(ip)s"),
{'resp': resp.status,
'ip': node['replication_ip']})
continue
#remote_hash 为 请求 'REPLICATE 返回的
remote_hash = pickle.loads(resp.read())
del resp
#找出本地后缀和远程后缀不同的
suffixes = [suffix for suffix in local_hash if
local_hash[suffix] !=
remote_hash.get(suffix, -1)]
#如果没有说明没有变动，则继续请求下一个节点
if not suffixes:
continue
#效果就是执行get_hashes方法
hashed, recalc_hash = tpool_reraise(
get_hashes,
job['path'], recalculate=suffixes,
reclaim_age=self.reclaim_age)
self.logger.update_stats('suffix.hashes', hashed)
local_hash = recalc_hash
#假如 local_hash 为 123 321 122 remote_hash 123 321 124 则 122为变化的
#文件路径hash值后三位会不会重复
suffixes = [suffix for suffix in local_hash if
local_hash[suffix] !=
remote_hash.get(suffix, -1)]
#找到了不同的并知道其节点则将其同步到对应的节点，是基于推送模式的，故传的数据是自己本地的数据
self.sync(node, job, suffixes) #同步变化的
with Timeout(self.http_timeout):
conn = http_connect(
node['replication_ip'], node['replication_port'],
node['device'], job['partition'], 'REPLICATE',
'/' + '-'.join(suffixes),
headers=self.headers)
conn.getresponse().read()
self.suffix_sync += len(suffixes)
self.logger.update_stats('suffix.syncs', len(suffixes))
except (Exception, Timeout):
self.logger.exception(_("Error syncing with node: %s") %
node)
#后缀数量 写日志时会用到
self.suffix_count += len(local_hash)
except (Exception, Timeout):
self.logger.exception(_("Error syncing partition"))
finally:
self.partition_times.append(time.time() - begin)
self.logger.timing_since('partition.update.timing', begin)update方法，中首先是获得本地文件中当前设备所对应hashes.pkl文件中每个后缀所对应的hahes值，形如{'a83': '0db7b416c9808517a1bb2157af20b09b'},其中key为文件内容hash值的后三字节，value为后缀文件夹下所有子文件夹下（即以文件内容的md5值为名字的文件夹）所有.data文件的文件名字的md5值,可以理解为所有文件名的md5值和。
hashed, local_hash = tpool_reraise(
get_hashes, job['path'],
do_listdir=(self.replication_count % 10) == 0,
reclaim_age=self.reclaim_age)def get_hashes(partition_dir, recalculate=None, do_listdir=False,
reclaim_age=ONE_WEEK):
"""
Get a list of hashes for the suffix dir. do_listdir causes it to mistrust
the hash cache for suffix existence at the (unexpectedly high) cost of a
listdir. reclaim_age is just passed on to hash_suffix.
:param partition_dir: absolute path of partition to get hashes for
:param recalculate: 形如 recalculate=['a83']
list of suffixes（后缀，即 hash值的后缀 310即为后缀 root@kinglion-Lenovo-Product:/srv/1/node/sdb1/objects/94238# ls
310 hashes.pkl ） which should be recalculated（重新计算） when got
:param do_listdir: force existence check for all hashes in the partition（对partion中的hashe强行执行检查）
:param reclaim_age: age at which to remove tombstones
:returns: tuple of (number of suffix dirs hashed, dictionary of hashes)
"""因没有传递recalulate这个参数故只有do_listdir为True时会强制执行重新计算后缀文件下所有文件名字的hash值。文件名字是时间戳，时间戳变了说明文件有更新，故需要和远程同步，检查是否为同一个版本，不是同一个版本的需要把本地版本传递给远程服务器。
attempts_left = len(job['nodes'])
#此时的nodes为除去本节点外的所有节点 因为 job['nodes]不包含本地节点get_more_nodes(int(job['partition']))能获得除去本partion所对应节点 外的其他所有节点
nodes = itertools.chain(
job['nodes'],
self.object_ring.get_more_nodes(int(job['partition'])))下边就是while循环，循环attempts_left次，
resp = http_connect(
node['replication_ip'], node['replication_port'],
node['device'], job['partition'], 'REPLICATE',
'', headers=self.headers).getresponse()根据迭代得到的node请求，因副本节点首先被迭代到，故首先请求副本节点。若果成功请求读取resp返回的内容，得到远程设备同一个partion下的remote_hash
suffixes = [suffix for suffix in local_hash if
local_hash[suffix] !=
remote_hash.get(suffix, -1)]
#如果没有说明没有变动，则继续请求下一个节点
if not suffixes:
continue执行同步：
self.sync(node, job, suffixes) #同步变化的 def sync(self, node, job, suffixes): # Just exists for doc anchor point
"""
Synchronize local suffix directories from a partition with a remote
node.
:param node: the "dev" entry for the remote node to sync with
:param job: information about the partition being synced
:param suffixes: a list of suffixes which need to be pushed
:returns: boolean indicating success or failure
"""
# self.sync_method = getattr(self, conf.get('sync_method') or 'rsync')
#配置没有 sync_method方法 则执行类自己的rsync方法
return self.sync_method(node, job, suffixes) self.sync_method = getattr(self, conf.get('sync_method') or 'rsync')
def rsync(self, node, job, suffixes):
"""
Uses rsync to implement the sync method. This was the first
sync method in Swift.
"""
if not os.path.exists(job['path']):
return False
args = [
'rsync',
'--recursive',
'--whole-file',
'--human-readable',
'--xattrs',
'--itemize-changes',
'--ignore-existing',
'--timeout=%s' % self.rsync_io_timeout,
'--contimeout=%s' % self.rsync_io_timeout,
'--bwlimit=%s' % self.rsync_bwlimit,
]
node_ip = rsync_ip(node['replication_ip'])
#包含了ip信息
if self.vm_test_mode:
rsync_module = '%s::object%s' % (node_ip, node['replication_port'])
else:
rsync_module = '%s::object' % node_ip
had_any = False
for suffix in suffixes:
spath = join(job['path'], suffix)
if os.path.exists(spath):
args.append(spath)
had_any = True
if not had_any:
return False
args.append(join(rsync_module, node['device'],
'objects', job['partition']))
#args里面包含了通的所有信息 包括设备名称，设备分区
return self._rsync(args) == 0 def _rsync(self, args):
"""
Execute the rsync binary to replicate a partition.
:returns: return code of rsync process. 0 is successful
"""
start_time = time.time()
ret_val = None
try:
with Timeout(self.rsync_timeout):
#此处即为同步操作了，推送模式
proc = subprocess.Popen(args,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
results = proc.stdout.read()
ret_val = proc.wait()
except Timeout:
self.logger.error(_("Killing long-running rsync: %s"), str(args))
proc.kill()
return 1 # failure response code
total_time = time.time() - start_time
for result in results.split('\n'):
if result == '':
continue
if result.startswith('cd+'):
continue
if not ret_val:
self.logger.info(result)
else:
self.logger.error(result)
if ret_val:
error_line = _('Bad rsync return code: %(ret)d <- %(args)s') % \
{'args': str(args), 'ret': ret_val}
if self.rsync_error_log_line_length:
error_line = error_line[:self.rsync_error_log_line_length]
self.logger.error(error_line)
elif results:
self.logger.info(
_("Successful rsync of %(src)s at %(dst)s (%(time).03f)"),
{'src': args[-2], 'dst': args[-1], 'time': total_time})
else:
self.logger.debug(
_("Successful rsync of %(src)s at %(dst)s (%(time).03f)"),
{'src': args[-2], 'dst': args[-1], 'time': total_time})
return ret_val #此处即为同步操作了，推送模式
proc = subprocess.Popen(args,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)若job[delete]为True出现这种情况的可能就是，因增删了设备，Ring 重新调整，当前partion中的备份不再有此服务器的ID如partion号为45678的在rebalance前的对于的备份设备的id为[1,2,3]，如果当前设备id为1，则重新rebalance后当前partion对应的备份为[4,2,3],则就会出现job[delete]为True的情况，我们看其代码具体实现：
def update_deleted(self, job):
"""
High-level method that replicates a single partition that doesn't
belong on (不应放在 )this node.
:param job: a dict containing info about the partition to be replicated
"""
#得到parition下对应的后缀
def tpool_get_suffixes(path):
return [suff for suff in os.listdir(path)
if len(suff) == 3 and isdir(join(path, suff))]
self.replication_count += 1
self.logger.increment('partition.delete.count.%s' % (job['device'],))
begin = time.time()
try:
responses = []
suffixes = tpool.execute(tpool_get_suffixes, job['path'])
if suffixes:
for node in job['nodes']:
success = self.sync(node, job, suffixes) #执行同步
if success:
with Timeout(self.http_timeout):
conn = http_connect(
node['replication_ip'],
node['replication_port'],
node['device'], job['partition'], 'REPLICATE',
'/' + '-'.join(suffixes), headers=self.headers)
conn.getresponse().read()
responses.append(success)
if self.handoff_delete:
# delete handoff if we have had handoff_delete successes
delete_handoff = len([resp for resp in responses if resp]) >= \
self.handoff_delete
else:
# delete handoff if all syncs were successful
delete_handoff = len(responses) == len(job['nodes']) and \
all(responses)
#suffixes为空或 请求的三个已经都响应成功后删除本地partion下的文件
if not suffixes or delete_handoff:
self.logger.info(_("Removing partition: %s"), job['path'])
tpool.execute(shutil.rmtree, job['path'], ignore_errors=True)
except (Exception, Timeout):
self.logger.exception(_("Error syncing handoff partition"))
finally:
self.partition_times.append(time.time() - begin)
self.logger.timing_since('partition.delete.timing', begin)
至此 replicate操作就讲解完毕,文中若有理解不合理之处，请指正，谢谢！
120
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
