ceph-deploy源码分析（一）——源码结构与cli <转>
ceph-deploy源码分析（一）——源码结构与cli
原文： http://www.hl10502.com/2017/06/15/ceph-deploy-cli/#more
ceph-deploy是部署Ceph集群的工具，可以通过SSH方式往远程主机上安装 Ceph 软件包、创建集群、增加监视器、收集（或销毁）密钥、增加 OSD 和元数据服务器、配置管理主机，甚至拆除集群。
ceph-deploy使用Python开发，GitHub为 https://github.com/ceph/ceph-deploy 。本次源码分析的ceph-deploy版本为1.5.37。
源码结构
ceph-deploy-1.5.37源码目录
├── ceph_deploy # ceph-deploy源码目录
├── ceph_deploy.egg-info # egg-info目录
├── CONTRIBUTING.rst # 贡献指南
├── LICENSE # MIT LICENSE
├── MANIFEST.in # 打包规则
├── PKG-INFO # PKG-INFO文件
├── README.rst # ceph-deploy介绍
├── scripts # 启动脚本目录
├── setup.cfg # setup.py配置
├── setup.py # ceph-deploy安装脚本
├── tox.ini # 标准化测试
└── vendor.py
ceph_deploy源码目录文件
├── admin.py # 子命令admin模块，将ceph.conf和client.admin key push到远程主机
├── calamari.py # 子命令calamari模块，连接calamari master
├── cli.py # CLI入口
├── cliutil.py # 为装饰器函数增加priority
├── conf # ceph.conf与cephdeploy.conf读取与写入相关操作目录
├── config.py # 子命令config模块，push ceph.conf文件到远程主机；从远程主机 pull ceph.conf文件
├── connection.py # 连接本地主机、远程主机
├── exc.py # 异常处理Error
├── forgetkeys.py # 子命令forgetkeys模块，本地移除authentication keys
├── gatherkeys.py # 子命令gatherkeys模块，从mon主机上拉取authentication keys
├── hosts # ceph-deploy在不同操作系统（centos、debian、fedora、rhel、suse）的操作
├── __init__.py # 初始化版本号信息，当前版本1.5.37
├── install.py # 子命令install模块，安装、卸载ceph包，清除数据
├── lib # vendor类库
├── mds.py # 子命令mds模块，mds管理
├── misc.py # 其他工具类，比如：mon host组装tuples
├── mon.py # 子命令mon模块，mon管理
├── new.py # 子命令new模块，部署集群
├── osd.py # 子命令osd模块，osd管理
├── pkg.py # 子命令pkg模块，逗号分隔的包安装、卸载
├── repo.py # 子命令repo模块，添加 yum repo
├── rgw.py # 子命令rgw模块，rgw管理
├── tests # 测试文件目录
├── util # util目录
└── validate.py # 参数校验函数
源码入口
script目录下的ceph-deploy文件是ceph-deploy的入口，安装之后是/usr/bin/ceph-deploy。
ceph-deploy的__main__函数调用ceph_deploy.cli的main函数
...
...
from ceph_deploy.cli import main
if __name__ == '__main__':
sys.exit(main())
cli模块
cli.py是命令行操作模块。
main函数调用_main函数
def main(args=None, namespace=None):
try:
_main(args=args, namespace=namespace)
finally:
# This block is crucial to avoid having issues with
# Python spitting non-sense thread exceptions. We have already
# handled what we could, so close stderr and stdout.
if not os.environ.get('CEPH_DEPLOY_TEST'):
try:
sys.stdout.close()
except:
pass
try:
sys.stderr.close()
except:
pass
_main函数
- 设置日志：Console Logger与File Logger添加到root_logger
- 调用argparse模块，解析cli参数
- 调用conf目录下的ceph-deploy模块set_overrides函数，从当前目录的cephdeploy.conf或~/.cephdeploy.conf文件获取ceph-deploy-global、ceph-deploy-[subcmd]配置项写入args
- 调用执行subcmd相应的模块
-
@catches((KeyboardInterrupt, RuntimeError, exc.DeployError,), handle_all=True) def _main(args=None, namespace=None): # Set console logging first with some defaults, to prevent having exceptions # before hitting logging configuration. The defaults can/will get overridden # later. # Console Logger # 命令行控制台日志 sh = logging.StreamHandler() # 不同级别的日志，使用不同的颜色区别：DEBUG蓝色；WARNIN黄色；ERROR红色；INFO白色 sh.setFormatter(log.color_format()) # 设置日志级别为WARNING sh.setLevel(logging.WARNING) # because we're in a module already, __name__ is not the ancestor of # the rest of the package; use the root as the logger for everyone # root_logger日志 root_logger = logging.getLogger() # allow all levels at root_logger, handlers control individual levels # 设置root_logger日志级别为DEBUG root_logger.setLevel(logging.DEBUG) # 将 sh添加到root_logger root_logger.addHandler(sh) # 获取解析cli的argparse，调用argparse模块 parser = get_parser() if len(sys.argv) < 2: parser.print_help() sys.exit() else: # 解析获取sys.argv中的ceph-deploy子命令和参数 args = parser.parse_args(args=args, namespace=namespace) # 设置日志级别 console_loglevel = logging.DEBUG # start at DEBUG for now if args.quiet: console_loglevel = logging.WARNING if args.verbose: console_loglevel = logging.DEBUG # Console Logger sh.setLevel(console_loglevel) # File Logger # 文件日志 fh = logging.FileHandler('ceph-deploy-{cluster}.log'.format(cluster=args.cluster)) fh.setLevel(logging.DEBUG) fh.setFormatter(logging.Formatter(log.FILE_FORMAT)) # 将 fh添加到root_logger root_logger.addHandler(fh) # Reads from the config file and sets values for the global # flags and the given sub-command # the one flag that will never work regardless of the config settings is # logging because we cannot set it before hand since the logging config is # not ready yet. This is the earliest we can do. # 从当前目录的cephdeploy.conf或~/.cephdeploy.conf文件获取ceph-deploy配置覆盖命令行参数 args = ceph_deploy.conf.cephdeploy.set_overrides(args) LOG.info("Invoked (%s): %s" % ( ceph_deploy.__version__, ' '.join(sys.argv)) ) log_flags(args) # args.func为cli中的subcmd子命令，调用相应的模块 return args.func(args)
get_parser函数
[ceph_deploy.cli]以以下方式配置：
模块名 = 模块包名:执行函数
比如： new = ceph_deploy.new:make
new作为ceph-deploy的子命令，执行ceph-deploy new命令时，执行make函数
其他的模块也类似：
mon = ceph_deploy.mon:make
osd = ceph_deploy.osd:make
rgw = ceph_deploy.rgw:make
mds = ceph_deploy.mds:make
config = ceph_deploy.config:make
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 | def get_parser(): # 调用argparse模块 parser = argparse.ArgumentParser( prog='ceph-deploy', formatter_class=argparse.RawDescriptionHelpFormatter, description='Easy Ceph deployment\n\n%s' % __header__, ) verbosity = parser.add_mutually_exclusive_group(required=False) verbosity.add_argument( '-v', '--verbose', action='store_true', dest='verbose', default=False, help='be more verbose', ) verbosity.add_argument( '-q', '--quiet', action='store_true', dest='quiet', help='be less verbose', ) parser.add_argument( '--version', action='version', version='%s' % ceph_deploy.__version__, help='the current installed version of ceph-deploy', ) parser.add_argument( '--username', help='the username to connect to the remote host', ) parser.add_argument( '--overwrite-conf', action='store_true', help='overwrite an existing conf file on remote host (if present)', ) parser.add_argument( '--cluster', metavar='NAME', help='name of the cluster', type=validate.alphanumeric, ) parser.add_argument( '--ceph-conf', dest='ceph_conf', help='use (or reuse) a given ceph.conf file', ) sub = parser.add_subparsers( title='commands', metavar='COMMAND', help='description', ) sub.required = True # 获取ceph_deploy.cli下的entry_points entry_points = [ (ep.name, ep.load()) for ep in pkg_resources.iter_entry_points('ceph_deploy.cli') ] # 根据priority排序 entry_points.sort( key=lambda name_fn: getattr(name_fn[1], 'priority', 100), ) # 将模块加入到子命令 for (name, fn) in entry_points: p = sub.add_parser( name, description=fn.__doc__, help=fn.__doc__, ) if not os.environ.get('CEPH_DEPLOY_TEST'): p.set_defaults(cd_conf=ceph_deploy.conf.cephdeploy.load()) # flag if the default release is being used p.set_defaults(default_release=False) fn(p) p.required = True parser.set_defaults( cluster='ceph', ) return parser |
argparse模块
cli命令的解析使用了argparse.py模块。argparse是Python标准库中命令行选项、参数和子命令的解析器，其是为替代已经过时的optparse模块，argparse在Python2.7中被引入。
argparse模块请参考
https://docs.python.org/2.7/library/argparse.html
http://python.usyiyi.cn/translate/python_278/library/argparse.html
set_overrides函数
conf目录下的ceph-deploy模块set_overrides函数
- 调用load()函数
- 判断ceph-deploy配置文件ceph-deploy-global、ceph-deploy-[subcommand]配置项，调用override_subcommand()函数写入args。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
def set_overrides(args, _conf=None):
"""
Read the configuration file and look for ceph-deploy sections
to set flags/defaults from the values found. This will alter the
``args`` object that is created by argparse.
"""
# Get the subcommand name to avoid overwritting values from other
# subcommands that are not going to be used
subcommand = args.func.__name__
command_section = 'ceph-deploy-%s' % subcommand
# 加载ceph-deploy配置
conf = _conf or load()
for section_name in conf.sections():
if section_name in ['ceph-deploy-global', command_section]:
# ceph-deploy-global、ceph-deploy-[subcommand]配置项写入args
override_subcommand(
section_name,
conf.items(section_name),
args
)
return args
load函数，调用location()函数
| 1 2 3 4 5 | def load(): parser = Conf() # 读取解析ceph-deploy配置文件 parser.read(location()) return parser |
location函数，调用_locate_or_create()函数
| 1 2 3 4 5 6 | def location(): """ Find and return the location of the ceph-deploy configuration file. If this file does not exist, create one in a default location. """ return _locate_or_create() |
_locate_or_create函数，判断当前目录的cephdeploy.conf或~/.cephdeploy.conf文件是否存在。
- 如果都不存在则调用create_stub函数创建一个~/.cephdeploy.conf文件。这个文件是根据模板创建的，内容为空。
- 如果存在（提前创建）cephdeploy.conf或~/.cephdeploy.conf文件，可以在文件中配置public_network、cluster_network、overwrite-conf等。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
def _locate_or_create():
home_config = path.expanduser('~/.cephdeploy.conf')
# With order of importance
locations = [
path.join(os.getcwd(), 'cephdeploy.conf'),
home_config,
]
for location in locations:
if path.exists(location):
logger.debug('found configuration file at: %s' % location)
return location
logger.info('could not find configuration file, will create one in $HOME')
create_stub(home_config)
return home_config