ceph-deploy源码分析（二）——new模块
原文：http://www.hl10502.com/2017/06/19/ceph-deploy-new/#more
ceph-deploy的new.py模块是用来开始部署新集群，创建ceph.conf、ceph.mon.keyring文件。
new 子命令格式如下
ceph-deploy new [-h] [--no-ssh-copykey] [--fsid FSID]
[--cluster-network CLUSTER_NETWORK]
[--public-network PUBLIC_NETWORK]
MON [MON ...]
部署集群
make函数priority为10，子命令设置的默认函数为new函数。
@priority(10)
def make(parser):
"""
Start deploying a new cluster, and write a CLUSTER.conf and keyring for it.
"""
parser.add_argument(
'mon',
metavar='MON',
nargs='+',
help='initial monitor hostname, fqdn, or hostname:fqdn pair',
type=arg_validators.Hostname(),
)
parser.add_argument(
'--no-ssh-copykey',
dest='ssh_copykey',
action='store_false',
default=True,
help='do not attempt to copy SSH keys',
)
parser.add_argument(
'--fsid',
dest='fsid',
help='provide an alternate FSID for ceph.conf generation',
)
parser.add_argument(
'--cluster-network',
help='specify the (internal) cluster network',
type=arg_validators.Subnet(),
)
parser.add_argument(
'--public-network',
help='specify the public network for a cluster',
type=arg_validators.Subnet(),
)
parser.set_defaults(
func=new,
)
部署新集群
new 函数开始部署新集群
- 创建ceph.conf文件,写入[global]fsid、mon_initial_members、mon_host、auth_cluster_required、auth_service_required、auth_client_required，如果参数中有public_network、cluster_network写入配置文件
- 调用new_mon_keyring函数创建ceph.mon.keyring文件
-
def new(args): if args.ceph_conf: raise RuntimeError('will not create a Ceph conf file if attemtping to re-use with `--ceph-conf` flag') LOG.debug('Creating new cluster named %s', args.cluster) # 生成配置 cfg = conf.ceph.CephConf() cfg.add_section('global') # 获取参数中的额fsid或者自动生成 fsid = args.fsid or uuid.uuid4() cfg.set('global', 'fsid', str(fsid)) # if networks were passed in, lets set them in the # global section if args.public_network: cfg.set('global', 'public network', str(args.public_network)) if args.cluster_network: cfg.set('global', 'cluster network', str(args.cluster_network)) # mon节点 mon_initial_members = [] # mon主机 mon_host = [] # 循环host for (name, host) in mon_hosts(args.mon): # Try to ensure we can ssh in properly before anything else # ssh key copy if args.ssh_copykey: ssh_copy_keys(host, args.username) # Now get the non-local IPs from the remote node # 连接远程主机 distro = hosts.get(host, username=args.username) # 获取主机的IP地址 remote_ips = net.ip_addresses(distro.conn) # custom cluster names on sysvinit hosts won't work if distro.init == 'sysvinit' and args.cluster != 'ceph': LOG.error('custom cluster names are not supported on sysvinit hosts') raise exc.ClusterNameError( 'host %s does not support custom cluster names' % host ) distro.conn.exit() # Validate subnets if we received any if args.public_network or args.cluster_network: # 校验IP地址 validate_host_ip(remote_ips, [args.public_network, args.cluster_network]) # Pick the IP that matches the public cluster (if we were told to do # so) otherwise pick the first, non-local IP LOG.debug('Resolving host %s', host) if args.public_network: ip = get_public_network_ip(remote_ips, args.public_network) else: ip = net.get_nonlocal_ip(host) LOG.debug('Monitor %s at %s', name, ip) mon_initial_members.append(name) try: socket.inet_pton(socket.AF_INET6, ip) mon_host.append("[" + ip + "]") LOG.info('Monitors are IPv6, binding Messenger traffic on IPv6') cfg.set('global', 'ms bind ipv6', 'true') except socket.error: mon_host.append(ip) LOG.debug('Monitor initial members are %s', mon_initial_members) LOG.debug('Monitor addrs are %s', mon_host) # mon_initial_members 有多个的话，中间用空格隔开 cfg.set('global', 'mon initial members', ', '.join(mon_initial_members)) # no spaces here, see http://tracker.newdream.net/issues/3145 # mon_host 有多个的话，中间没有空格 cfg.set('global', 'mon host', ','.join(mon_host)) # override undesirable defaults, needed until bobtail # http://tracker.ceph.com/issues/6788 cfg.set('global', 'auth cluster required', 'cephx') cfg.set('global', 'auth service required', 'cephx') cfg.set('global', 'auth client required', 'cephx') path = '{name}.conf'.format( name=args.cluster, ) # 创建mon keyring new_mon_keyring(args) LOG.debug('Writing initial config to %s...', path) tmp = '%s.tmp' % path with open(tmp, 'w') as f: # 保存ceph配置文件 cfg.write(f) try: os.rename(tmp, path) except OSError as e: if e.errno == errno.EEXIST: raise exc.ClusterExistsError(path) else: raise
注意：
mon_initial_members 有多个的话，中间用空格隔开
mon_host 有多个的话，中间没有空格
创建ceph.mon.keyring文件
new_mon_keyring函数创建ceph.mon.keyring文件
def new_mon_keyring(args):
LOG.debug('Creating a random mon key...')
mon_keyring = '[mon.]\nkey = %s\ncaps mon = allow *\n' % generate_auth_key()
keypath = '{name}.mon.keyring'.format(
name=args.cluster,
)
oldmask = os.umask(0o77)
LOG.debug('Writing monitor keyring to %s...', keypath)
try:
tmp = '%s.tmp' % keypath
with open(tmp, 'w', 0o600) as f:
f.write(mon_keyring)
try:
os.rename(tmp, keypath)
except OSError as e:
if e.errno == errno.EEXIST:
raise exc.ClusterExistsError(keypath)
else:
raise
finally:
os.umask(oldmask)
手工部署集群
以ceph-deploy部署集群：ceph-deploy new ceph-231为例，对应的手工操作。
获取ip地址
执行以下命令，通过正则表达式获取IP地址192.168.217.231
[root@ceph-231 ceph-cluster]# /usr/sbin/ip link show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master ovs-system state UP mode DEFAULT qlen 1000
link/ether 02:03:e7:fc:dc:36 brd ff:ff:ff:ff:ff:ff
3: ovs-system: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT
link/ether 86:f4:14:e3:1b:b2 brd ff:ff:ff:ff:ff:ff
4: xenbr0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN mode DEFAULT
link/ether 02:03:e7:fc:dc:36 brd ff:ff:ff:ff:ff:ff
[root@ceph-231 ceph-cluster]# /usr/sbin/ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master ovs-system state UP qlen 1000
link/ether 02:03:e7:fc:dc:36 brd ff:ff:ff:ff:ff:ff
3: ovs-system: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN
link/ether 86:f4:14:e3:1b:b2 brd ff:ff:ff:ff:ff:ff
4: xenbr0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN
link/ether 02:03:e7:fc:dc:36 brd ff:ff:ff:ff:ff:ff
inet 192.168.217.231/24 brd 192.168.217.255 scope global xenbr0
valid_lft forever preferred_lft forever
创建ceph.conf
[root@ceph-231 ceph-cluster]# vi ceph.conf
[global]
fsid = a3b9b0aa-01ab-4e1b-bba3-6f5317b0795b
mon_initial_members = ceph-231
mon_host = 192.168.217.231
auth_cluster_required = cephx
auth_service_required = cephx
auth_client_required = cephx
public_network = 192.168.217.231
创建ceph.mon.keyring
可以通过ceph-authtool命令生成
[root@ceph-231 ceph-cluster]# ceph-authtool --create-keyring /tmp/ceph.mon.keyring --gen-key -n mon. --cap mon 'allow *'
creating /tmp/ceph.mon.keyring
[root@ceph-231 ~]# cat /tmp/ceph.mon.keyring
[mon.]
key = AQCzxEhZC7tICxAAuHK5GipD96enMuhv82CCLg==
caps mon = "allow *"将/tmp/ceph.mon.keyring内容复制到ceph.mon.keyring
[root@ceph-231 ceph-cluster]# vi ceph.mon.keyring
[mon.]
key = AQCzxEhZC7tICxAAuHK5GipD96enMuhv82CCLg==
caps mon = allow
243
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
