def ping(host=None, port=None, all=False):
"""Send ping command
:param all: If true, send command to all
:param host: host info for redis
:param port: port info for redis
"""
if not isinstance(all, bool):
msg = message.get('error_option_type_not_boolean').format(option='all')
logger.error(msg)
return
if (not host or not port) and not all:
msg = message.get('use_host_port_or_option_all')
logger.error(msg)
return
if all:
meta = []
ret = RedisCliUtil.command_all_async('ping 2>&1')
pong_cnt = 0
for m_s, host, port, result, _ in ret:
addr = '{}:{}'.format(host, port)
if result == 'OK':
pong_cnt += 1
else:
meta.append([m_s, addr, color.red('FAIL')])
if meta:
utils.print_table([['TYPE', 'ADDR', 'RESULT']] + meta)
msg = message.get('counting_alive_redis')
msg = msg.format(alive=pong_cnt, total=len(ret))
logger.info(msg)
return
if host and port:
_command('ping', False, host, port)
def __init__(self, exit_status, command, hostname, port):
self.exit_status = exit_status
self.command = command
self.hostname = hostname
self.port = port
msg = [
'CommandError:', 'exit_status={}'.format(exit_status),
'command={}'.format(command),
'host:port={}:{}'.format(hostname, port)
]
logger.error('\n'.join(msg))
def base_directory(default='~/tsr2'):
logger.debug('ask base directory')
result = ask(message.get('ask_base_directory'), default=default)
if not result.startswith(('~', '/')):
logger.error(message.get('error_invalid_path').format(value=result))
return base_directory()
logger.info('OK, {}'.format(result))
cli_config = config.get_cli_config()
cli_config['base_directory'] = result
config.save_cli_config(cli_config)
return result
def reset_oom(all=False, host=None, port=0):
"""Send reset oom command
:param all: If true, send command to all
:param host: host info for redis
:param port: port info for redis
"""
if not isinstance(all, bool):
msg = message.get('error_option_type_not_boolean').format(option='all')
logger.error(msg)
return
sub_cmd = 'resetOom'
_command(sub_cmd, all, host, port)
def set(self, key, value, all=False, save=False, host=None, port=None):
"""Command: redis-cli config set
:param key: target key
:param value: value to set
:param save: If true, save value to config file
:param all: If true, send command to all redis
:param host: host info for redis
:param port: port info for redis
"""
if not isinstance(all, bool):
msg = m.get('error_option_type_not_boolean')
msg = msg.format(options='all')
logger.error(msg)
return
if not isinstance(save, bool):
msg = m.get('error_option_type_not_boolean')
msg = msg.format(options='save')
logger.error(msg)
return
if (not host or not port) and not all:
msg = m.get('use_host_port_or_option_all')
logger.error(msg)
return
sub_cmd = 'config set {key} {value} 2>&1'.format(key=key, value=value)
if all:
meta = []
ret = RedisCliUtil.command_all_async(sub_cmd)
ok_cnt = 0
for m_s, host, port, result, message in ret:
addr = '{}:{}'.format(host, port)
if result == 'OK':
if utils.to_str(message) == 'OK':
ok_cnt += 1
else:
meta.append([m_s, addr, color.red(message)])
else:
meta.append([m_s, addr, color.red('FAIL')])
if meta:
utils.print_table([['TYPE', 'ADDR', 'RESULT']] + meta)
logger.info('success {}/{}'.format(ok_cnt, len(ret)))
else:
output = RedisCliUtil.command(sub_cmd=sub_cmd,
host=host,
port=port,
formatter=self.no_print)
output = output.strip()
if output == "OK":
logger.info(output)
else:
logger.error(output)
if save:
RedisCliUtil.save_redis_template_config(key, value)
center = Center()
center.update_ip_port()
success = center.check_hosts_connection()
if not success:
return
center.configure_redis()
center.sync_conf()
def metakeys(key, all=False, host=None, port=0):
"""Get metakeys
:param key: target key
:param all: If true, send command to all
:param host: host info for redis
:param port: port info for redis
"""
if not isinstance(all, bool):
msg = message.get('error_option_type_not_boolean').format(option='all')
logger.error(msg)
return
sub_cmd = 'metakeys "%s"' % key
_command(sub_cmd, all, host, port)
def reset_info(key, all=False, host=None, port=0):
"""Send reset info
:param key: resetting target key string
:param all: If true, send command to all
:param host: host info for redis
:param port: port info for redis
"""
if not isinstance(all, bool):
msg = message.get('error_option_type_not_boolean').format(option='all')
logger.error(msg)
return
sub_cmd = 'resetInfo %s' % key
_command(sub_cmd, all, host, port)
def ssd_count(save, default=None):
logger.debug('ask ssd count')
deploy_history = config.get_deploy_history()
if not default:
default = deploy_history['ssd_count']
result = int(askInt(message.get('ask_ssd_count'), default=str(default)))
if result <= 0:
logger.error(message.get('error_ssd_count_less_than_1'))
return ssd_count(save=save, default=default)
if save:
deploy_history['ssd_count'] = result
config.save_deploy_history(deploy_history)
logger.info('OK, {}'.format(result))
return result
def replicas(save, default=None):
logger.debug('ask replicas')
deploy_history = config.get_deploy_history()
if not default:
default = deploy_history['replicas']
result = askInt(message.get('ask_replicas'), default=str(default))
result = int(result)
if result < 0:
logger.error(message.get('error_replicas_less_than_0'))
return replicas(save, default=default)
if save:
deploy_history['replicas'] = result
config.save_deploy_history(deploy_history)
logger.info('OK, {}'.format(result))
return result
def host_for_monitor(host_list):
formatted = []
for i, v in enumerate(host_list):
formatted.append(' ({}) {}'.format(i + 1, v))
stringfied_list = '\n'.join(formatted)
msg = '\n'.join(message.get('ask_host_for_monitor'))
msg = msg.format(list=stringfied_list)
target_num = int(askInt(msg, default='1'))
while True:
if target_num > 0 and target_num <= len(host_list):
break
msg = message.get('error_select_number').format(
max_number=len(host_list))
logger.error(msg)
target_num = int(askInt(''))
return host_list[target_num - 1]
def clean(self, logs=False):
"""Clean cluster
Delete redis config, data, node configuration.
:param log: Delete log of redis
"""
if not isinstance(logs, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='logs')
logger.error(msg)
return
center = Center()
center.update_ip_port()
if logs:
center.remove_all_of_redis_log_force()
return
center.cluster_clean()
msg = message.get('apply_after_restart')
logger.info(msg)
def run_sync(host=None):
"""Import clusters from the host
"""
if host is None:
logger.error('host information is not available')
return None
cluster_base = config.get_base_directory()
if not os.path.exists(cluster_base):
logger.error('cluster does not exist on the localhost.')
os.mkdir(cluster_base)
cluster_set = set(
filter(lambda x: re.match(r'cluster_[\d]+', x),
os.listdir(cluster_base)))
client = net.get_ssh(host)
if not net.is_dir(client, cluster_base):
logger.error('cluster does not exist on the host({}).'.format(host))
return None
target_cluster_set = set(
filter(
lambda x: re.match(r'cluster_[\d]+', x),
net.ssh_execute(client, 'ls {}'.format(cluster_base))[1].split()))
conflict_cluster = cluster_set & target_cluster_set
import_target = (cluster_set ^ target_cluster_set) & target_cluster_set
for cluster in conflict_cluster:
msg = message.get('ask_cluster_overwrite').format(
cluster=" ".join(cluster.split('_')))
overwrite = ask_util.askBool(msg, default='n')
if overwrite:
import_target.add(cluster)
os.system("rm -rf {}".format(cluster_base + "/" + cluster))
for target in import_target:
os.system("rsync -a {} {}".format(
host + ":" + cluster_base + "/" + target, cluster_base))
logger.info("Importing cluster complete...")
def stop(self, force=False, master=True, slave=True):
"""Stop cluster
:param force: Force the cluster to shut down
:param master: If exclude master cluster, set False
:param slave: If exclude slave cluster, set False
"""
if not isinstance(force, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='force')
logger.error(msg)
return
if not isinstance(master, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='master')
logger.error(msg)
return
if not isinstance(slave, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='slave')
logger.error(msg)
return
center = Center()
center.update_ip_port()
success = center.check_hosts_connection()
if not success:
return
center.stop_redis(force, master=master, slave=slave)
def _handle(text):
if text == '':
return
if text == 'clear':
utils.clear_screen()
return
text = text.replace('-- --help', '?')
text = text.replace('--help', '?')
text = text.replace('?', '-- --help')
err_flg = True
try:
fire.Fire(component=Command, command=text)
err_flg = False
except KeyboardInterrupt:
msg = message.get('cancel_command_input')
logger.warning('\b\b' + msg)
except KeyError as ex:
logger.warn('[%s] command fail' % text)
logger.exception(ex)
except TypeError as ex:
logger.exception(ex)
except IOError as ex:
if ex.errno == 2:
msg = message.get('error_file_not_exist').format(file=ex.filename)
logger.error(msg)
else:
logger.exception(ex)
except EOFError:
msg = message.get('cancel_command_input')
logger.warning('\b\b' + msg)
except utils.CommandError as ex:
logger.exception(ex)
except FireError:
pass
except FireExit:
pass
except (
HostNameError,
HostConnectionError,
SSHConnectionError,
FileNotExistError,
YamlSyntaxError,
PropsSyntaxError,
PropsKeyError,
PropsError,
SSHCommandError,
ClusterRedisError,
ClusterNotExistError,
ClusterIdError,
EnvError,
) as ex:
logger.error('{}: {}'.format(ex.class_name(), str(ex)))
except LightningDBError as ex:
logger.error('[ErrorCode {}] {}'.format(ex.error_code, str(ex)))
except BaseException as ex:
logger.exception(ex)
finally:
return err_flg
def _initial_check():
try:
# Simple check to see if ssh access to localhost is possible
net.get_ssh('localhost')
except paramiko.ssh_exception.SSHException:
msg = message.get('error_ssh_connection').format(host='localhost')
logger.error(msg)
exit(1)
cli_config = config.get_cli_config()
try:
base_directory = cli_config['base_directory']
except KeyError:
pass
except TypeError:
root_of_cli_config = config.get_root_of_cli_config()
conf_path = os.path.join(root_of_cli_config, 'config')
os.system('rm {}'.format(conf_path))
base_directory = None
if not base_directory or not base_directory.startswith(('~', '/')):
base_directory = ask_util.base_directory()
base_directory = os.path.expanduser(base_directory)
if not os.path.isdir(base_directory):
os.system('mkdir -p {}'.format(base_directory))
def start(self, profile=False, master=True, slave=True):
"""Start cluster
:param master: If exclude master cluster, set False
:param slave: If exclude slave cluster, set False
"""
logger.debug("command 'cluster start'")
if not isinstance(profile, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='profile')
logger.error(msg)
return
if not isinstance(master, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='master')
logger.error(msg)
return
if not isinstance(slave, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='slave')
logger.error(msg)
return
center = Center()
center.update_ip_port()
success = center.check_hosts_connection()
if not success:
return
center.ensure_cluster_exist()
if master:
master_alive_count = center.get_alive_master_redis_count()
master_alive_count_mine = center.get_alive_master_redis_count(
check_owner=True
)
not_mine_count = master_alive_count - master_alive_count_mine
if not_mine_count > 0:
msg = message.get('error_cluster_start_master_collision')
msg = '\n'.join(msg).format(count=not_mine_count)
raise LightningDBError(11, msg)
if slave:
slave_alive_count = center.get_alive_slave_redis_count()
slave_alive_count_mine = center.get_alive_slave_redis_count(
check_owner=True
)
not_mine_count = slave_alive_count - slave_alive_count_mine
if not_mine_count > 0:
msg = message.get('error_cluster_start_slave_collision')
msg = '\n'.join(msg).format(count=not_mine_count)
raise LightningDBError(12, msg)
center.backup_server_logs(master=master, slave=slave)
center.create_redis_data_directory()
# equal to cluster.configure()
center.configure_redis()
center.sync_conf(show_result=True)
center.start_redis_process(profile, master=master, slave=slave)
center.wait_until_all_redis_process_up(master=master, slave=slave)
def run_deploy(cluster_id=None,
history_save=True,
clean=False,
strategy="none"):
"""Install LightningDB package.
:param cluster_id: cluster id
:param history_save: save input history and use as default
:param clean: delete redis log, node configuration
:param strategy:
none(default): normal deploy,
zero-downtime: re-deploy without stop
"""
# validate cluster id
if cluster_id is None:
cluster_id = config.get_cur_cluster_id(allow_empty_id=True)
if cluster_id < 0:
msg = message.get('error_invalid_cluster_on_deploy')
logger.error(msg)
return
if not cluster_util.validate_id(cluster_id):
raise ClusterIdError(cluster_id)
# validate option
if not isinstance(history_save, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='history-save')
logger.error(msg)
return
logger.debug("option '--history-save': {}".format(history_save))
if not isinstance(clean, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='clean')
logger.error(msg)
return
logger.debug("option '--clean': {}".format(clean))
strategy_list = ["none", "zero-downtime"]
if strategy not in strategy_list:
msg = message.get('error_deploy_strategy').format(value=strategy,
list=strategy_list)
logger.error(msg)
return
if strategy == "zero-downtime":
run_cluster_use(cluster_id)
_deploy_zero_downtime(cluster_id)
return
_deploy(cluster_id, history_save, clean)
def get(self, key, all=False, host=None, port=None):
"""Command: redis-cli config get
:param key: redis config keyword
:param all: If true, send command to all redis
:param host: host info for redis
:param port: port info for redis
"""
if not isinstance(all, bool):
msg = m.get('error_option_type_not_boolean')
msg = msg.format(options='all')
logger.error(msg)
return
if (not host or not port) and not all:
msg = m.get('use_host_port_or_option_all')
logger.error(msg)
return
sub_cmd = 'config get "{key}" 2>&1'.format(key=key)
if all:
meta = []
ret = RedisCliUtil.command_all_async(sub_cmd)
for m_s, host, port, result, message in ret:
addr = '{}:{}'.format(host, port)
if result == 'OK':
if message:
_, value = message.split('\n')
meta.append([m_s, addr, value])
else:
meta.append([m_s, addr, color.red('Invalid Key')])
else:
meta.append([m_s, addr, color.red(result)])
utils.print_table([['TYPE', 'ADDR', 'RESULT']] + meta)
else:
output = RedisCliUtil.command(sub_cmd=sub_cmd,
host=host,
port=port,
formatter=self.no_print)
output = output.strip()
if output:
key, value = output.split('\n')
logger.info(value)
else:
msg = m.get('error_invalid_key').format(key=key)
logger.error(msg)
def run_monitor(n=10, t=2):
"""Monitoring logs of redis.
:param n: number of lines to print log
:param t: renewal cycle(sec)
"""
if not isinstance(n, int):
msg = message.get('error_option_type_not_number').format(option='n')
logger.error(msg)
return
if not isinstance(t, int) and not isinstance(t, float):
msg = message.get('error_option_type_not_float').format(option='t')
logger.error(msg)
return
try:
sp.check_output('which tail', shell=True)
except Exception:
msg = message.get('error_not_found_command_tail')
logger.error(msg)
return
cluster_id = config.get_cur_cluster_id()
path_of_fb = config.get_path_of_fb(cluster_id)
sr2_redis_log = path_of_fb['sr2_redis_log']
log_files = '{}/servers*'.format(sr2_redis_log)
host_list = config.get_master_host_list()
target_host = ask_util.host_for_monitor(host_list)
try:
sp.check_output('which watch', shell=True)
command = "ssh -t {} watch -n {} 'tail -n {} {}'".format(
target_host, t, n, log_files)
sp.call(command, shell=True)
except Exception:
msg = message.get('error_not_found_command_watch')
logger.warning(msg)
logger.info(message.get('message_for_exit'))
command = "tail -F -s {} {}".format(t, log_files)
client = net.get_ssh(target_host)
net.ssh_execute_async(client, command)
def master_ports(save, cluster_id, default_count=None):
logger.debug('ask master ports')
deploy_history = config.get_deploy_history()
if not default_count:
default_count = deploy_history['master_count']
msg = message.get('ask_master_count')
m_count = int(askInt(msg, default=str(default_count)))
if m_count <= 0:
logger.error(message.get('error_master_count_less_than_1'))
return master_ports(cluster_id, default_count)
logger.info('OK, {}'.format(m_count))
if save:
deploy_history['master_count'] = m_count
config.save_deploy_history(deploy_history)
start_m_ports = START_PORT + cluster_id * MASTER_OFFSET
end_m_ports = start_m_ports + m_count - 1
if start_m_ports == end_m_ports:
default_m_ports = str(start_m_ports)
else:
default_m_ports = '{}-{}'.format(start_m_ports, end_m_ports)
while True:
result = ask(message.get('ask_ports'), default=default_m_ports)
result = list(map(lambda x: x.strip(), result.split(',')))
valid = True
m_ports = set()
pattern = re.compile('[0-9]+-[0-9]+')
for item in result:
# range number
matched = pattern.match(item)
if matched:
s, e = map(int, item.split('-'))
if s > e:
msg = message.get('error_invalid_range').format(value=item)
logger.error(msg)
valid = False
break
m_ports.update(range(s, e + 1))
continue
# single number
elif utils.is_number(item):
m_ports.add(int(item))
continue
else:
msg = message.get('error_invalid_input').format(value=item)
logger.error(msg)
valid = False
break
if not valid:
continue
out_of_range = []
for port in m_ports:
if not port_range_safe(port):
out_of_range.append(port)
if out_of_range:
msg = message.get('error_port_range').format(
minimum=PORT_MININUM,
maximum=PORT_MAXIMUM,
value=out_of_range,
)
logger.error(msg)
continue
if valid and len(m_ports) != m_count:
msg = message.get('error_port_count_different').format(
advance=m_count,
current=len(m_ports),
)
logger.error(msg)
continue
if valid:
break
m_ports = sorted(list(m_ports))
logger.info('OK, {}'.format(result))
return m_ports
def restart(
self,
force_stop=False,
reset=False,
cluster=False,
profile=False,
yes=False,
):
"""Restart cluster
:param force_stop: Force the cluster to shuto down
:param reset: Delete redis config, data, node configuration
:param cluster: Create cluster after cluster start
:param yes: Skip confirm information when cluster create
"""
if not isinstance(force_stop, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='force-stop')
logger.error(msg)
return
if not isinstance(reset, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='reset')
logger.error(msg)
return
if not reset and cluster:
msg = message.get('error_option_use_with')
msg = msg.format(option='cluster', with_option='reset')
logger.error(msg)
return
if not isinstance(cluster, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='cluster')
logger.error(msg)
return
if not cluster and yes:
msg = message.get('error_option_use_with')
msg = msg.format(option='yes', with_option='cluster')
msg = "option '--yes' can used only with option '--cluster'"
logger.error(msg)
return
if not isinstance(yes, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='yes')
logger.error(msg)
return
center = Center()
center.update_ip_port()
success = center.check_hosts_connection()
if not success:
return
center.stop_redis(force=force_stop)
if reset:
self.clean()
self.start(profile=profile)
if cluster:
self.create(yes=yes)
def add_slave(self, yes=False):
"""Add slave of cluster
Add slaves to cluster that configured master only.
:param yes: Skip confirm information
"""
logger.debug('add_slave')
if not isinstance(yes, bool):
msg = message.get('error_option_type_not_boolean')
msg = msg.format(option='yes')
logger.error(msg)
return
center = Center()
center.update_ip_port()
# check
s_hosts = center.slave_host_list
s_ports = center.slave_port_list
if not s_hosts:
msg = message.get('error_slave_host_empty')
raise ClusterRedisError(msg)
if not s_ports:
msg = message.get('error_slave_port_empty')
raise ClusterRedisError(msg)
success = center.check_hosts_connection(hosts=s_hosts)
if not success:
return
center.ensure_cluster_exist()
slave_alive_count = center.get_alive_slave_redis_count()
slave_alive_count_mine = center.get_alive_slave_redis_count(
check_owner=True
)
not_mine_count = slave_alive_count - slave_alive_count_mine
if not_mine_count > 0:
msg = message.get('error_cluster_start_slave_collision')
msg = '\n'.join(msg).format(count=not_mine_count)
raise LightningDBError(12, msg)
# confirm info
result = center.confirm_node_port_info(skip=yes)
if not result:
msg = message.get('cancel')
logger.warning(msg)
return
# clean
center.cluster_clean(master=False)
# backup logs
center.backup_server_logs(master=False)
center.create_redis_data_directory(master=False)
# configure
center.configure_redis(master=False)
center.sync_conf()
# start
center.start_redis_process(master=False)
center.wait_until_all_redis_process_up()
# change redis config temporarily
key = 'cluster-node-timeout'
origin_s_value = center.cli_config_get(key, s_hosts[0], s_ports[0])
if not origin_s_value:
msg = "RedisConfigKeyError: '{}'".format(key)
logger.warning(msg)
if origin_s_value:
# cli config set cluster-node-timeout 2000
logger.debug('set cluster node time out 2000 for create')
center.cli_config_set_all(key, '2000', s_hosts, s_ports)
# create
center.replicate()
if origin_s_value:
# cli config restore cluster-node-timeout
logger.debug('restore cluster node time out')
center.cli_config_set_all(key, origin_s_value, s_hosts, s_ports)
def restore(self, cluster_id, tag=None):
"""Restore cluster
:param cluster_id: target cluster id
:param tag: Tag of backup, if omitted, restore the most recent backup file
"""
logger.debug('cluster restore: cluster_id={}, tag={}'.format(
cluster_id,
tag
))
if not cluster_util.validate_id(cluster_id):
raise ClusterIdError(cluster_id)
# find restore folder with tag (local)
path_of_fb = config.get_path_of_fb(cluster_id)
cluster_backup_path = path_of_fb['cluster_backup_path']
if tag is None:
backup_list = os.listdir(cluster_backup_path)
pattern = 'cluster_{}_bak_'.format(cluster_id)
filtered = filter(lambda x: x.startswith(pattern), backup_list)
sorted_list = sorted(list(filtered))
if not sorted_list:
msg = message.get('error_not_found_any_backup')
logger.error('BackupNotExistError: ' + msg)
return
tag = sorted_list[-1]
logger.debug("tag option is empty, auto select: {}".format(tag))
cluster_restore_dir = tag
backup_path = os.path.join(cluster_backup_path, cluster_restore_dir)
if not os.path.isdir(backup_path):
msg = message.get('error_not_found_backup').format(tag=tag)
logger.error('BackupNotExistError: ' + msg)
return
# get hosts from cluster props
props_path = os.path.join(
backup_path,
'tsr2-assembly-1.0.0-SNAPSHOT',
'conf',
'redis.properties'
)
hosts = config.get_props(props_path, 'sr2_redis_master_hosts', [])
# check status of hosts
success = Center().check_hosts_connection(hosts, True)
if not success:
msg = message.get('error_exist_unavailable_host')
logger.error(msg)
return
logger.debug('Connection of all hosts ok.')
success = Center().check_include_localhost(hosts)
if not success:
msg = message.get('error_not_include_localhost')
logger.error(msg)
return
# check all host tag folder: OK / NOT FOUND
msg = message.get('check_backup_info')
logger.info(msg)
buf = []
for host in hosts:
client = net.get_ssh(host)
if not net.is_dir(client, backup_path):
logger.debug('cannot find backup dir: {}-{}'.format(
host,
cluster_restore_dir
))
buf.append([host, color.red('NOT FOUND')])
client.close()
if buf:
utils.print_table([['HOST', 'RESULT'] + buf])
return
logger.info('OK')
# backup cluster
new_tag = time.strftime("%Y%m%d%H%M%S", time.gmtime())
cluster_backup_dir = 'cluster_{}_bak_{}'.format(cluster_id, new_tag)
for host in hosts:
Center().cluster_backup(host, cluster_id, cluster_backup_dir)
# restore cluster
command = "cp -r {} {}/cluster_{}".format(
backup_path,
path_of_fb['base_directory'],
cluster_id
)
for host in hosts:
msg = message.get('restore_cluster')
msg = msg.format(tag=cluster_backup_dir, host=host)
logger.info(msg)
client = net.get_ssh(host)
net.ssh_execute(client, command)
client.close()
logger.info("OK")
def _deploy(cluster_id, history_save, clean):
deploy_state = DeployUtil().get_state(cluster_id)
if deploy_state == DEPLOYED:
msg = message.get('ask_deploy_again')
msg = msg.format(cluster_id=cluster_id)
msg = color.yellow(msg)
yes = ask_util.askBool(msg, default='n')
if not yes:
logger.info(message.get('cancel'))
return
restore_yes = None
no_localhost = False
current_time = time.strftime("%Y%m%d%H%M%S", time.gmtime())
cluster_backup_dir = 'cluster_{}_bak_{}'.format(cluster_id, current_time)
conf_backup_dir = 'cluster_{}_conf_bak_{}'.format(cluster_id, current_time)
tmp_backup_dir = 'cluster_{}_conf_bak_{}'.format(cluster_id, 'tmp')
meta = [['NAME', 'VALUE']]
path_of_fb = config.get_path_of_fb(cluster_id)
conf_path = path_of_fb['conf_path']
props_path = path_of_fb['redis_properties']
cluster_path = path_of_fb['cluster_path']
path_of_cli = config.get_path_of_cli(cluster_id)
conf_backup_path = path_of_cli['conf_backup_path']
tmp_backup_path = os.path.join(conf_backup_path, tmp_backup_dir)
local_ip = config.get_local_ip()
# ask installer
installer_path = ask_util.installer()
installer_name = os.path.basename(installer_path)
meta.append(['installer', installer_name])
# ask restore conf
if deploy_state == DEPLOYED:
restore_yes = ask_util.askBool(message.get('ask_restore_conf'))
meta.append(['restore', restore_yes])
# input props
hosts = []
if deploy_state == DEPLOYED:
if restore_yes:
meta += DeployUtil().get_meta_from_props(props_path)
hosts = config.get_props(props_path, 'sr2_redis_master_hosts')
else:
if not os.path.isdir(conf_backup_path):
os.mkdir(conf_backup_path)
if os.path.exists(tmp_backup_path):
msg = message.get('ask_load_history_of_previous_modification')
yes = ask_util.askBool(msg)
if not yes:
shutil.rmtree(tmp_backup_path)
if not os.path.exists(tmp_backup_path):
os.mkdir(tmp_backup_path)
shutil.copy(os.path.join(conf_path, 'redis.properties'),
os.path.join(tmp_backup_path, 'redis.properties'))
tmp_props_path = os.path.join(tmp_backup_path, 'redis.properties')
editor.edit(tmp_props_path, syntax='sh')
meta += DeployUtil().get_meta_from_props(tmp_props_path)
hosts = config.get_props(tmp_props_path, 'sr2_redis_master_hosts')
else:
# new deploy
props_dict = ask_util.props(cluster_id, save=history_save)
hosts = props_dict['hosts']
meta += DeployUtil().get_meta_from_dict(props_dict)
utils.print_table(meta)
msg = message.get('confirm_deploy_information')
yes = ask_util.askBool(msg)
if not yes:
logger.info(message.get('cancel'))
return
# check node status
success = Center().check_hosts_connection(hosts, True)
if not success:
msg = message.get('error_exist_unavailable_host')
logger.error(msg)
return
logger.debug('Connection of all hosts ok.')
success = Center().check_include_localhost(hosts)
if not success:
no_localhost = True
# get port info
if deploy_state == DEPLOYED:
if restore_yes:
key = 'sr2_redis_master_ports'
m_ports = config.get_props(props_path, key, [])
key = 'sr2_redis_slave_ports'
s_ports = config.get_props(props_path, key, [])
replicas = len(s_ports) // len(m_ports)
else:
key = 'sr2_redis_master_ports'
m_ports = config.get_props(tmp_props_path, key, [])
key = 'sr2_redis_slave_ports'
s_ports = config.get_props(tmp_props_path, key, [])
replicas = len(s_ports) // len(m_ports)
else:
m_ports = props_dict['master_ports']
s_ports = props_dict['slave_ports']
replicas = props_dict['replicas']
while True:
msg = message.get('check_port')
logger.info(msg)
host_ports_list = []
for host in hosts:
host_ports_list.append((host, m_ports + s_ports))
conflict = Center().check_port_is_enable(host_ports_list)
if not conflict:
logger.info("OK")
break
utils.print_table([["HOST", "PORT"]] + conflict)
msg = message.get('ask_port_collision')
msg = color.yellow(msg)
yes = ask_util.askBool(msg)
if yes:
logger.info("OK")
break
m_ports = ask_util.master_ports(False, cluster_id)
replicas = ask_util.replicas(False)
s_ports = ask_util.slave_ports(cluster_id, len(m_ports), replicas)
if deploy_state == DEPLOYED:
if restore_yes:
key = 'sr2_redis_master_ports'
value = cluster_util.convert_list_2_seq(m_ports)
config.set_props(props_path, key, value)
key = 'sr2_redis_slave_ports'
value = cluster_util.convert_list_2_seq(s_ports)
config.set_props(props_path, key, value)
else:
key = 'sr2_redis_master_ports'
value = cluster_util.convert_list_2_seq(m_ports)
config.set_props(tmp_props_path, key, value)
key = 'sr2_redis_slave_ports'
value = cluster_util.convert_list_2_seq(s_ports)
config.set_props(tmp_props_path, key, value)
else:
props_dict['master_ports'] = m_ports
props_dict['slave_ports'] = s_ports
props_dict['replicas'] = replicas
# if pending, delete legacy on each hosts
if no_localhost:
if DeployUtil().get_state(cluster_id, local_ip) == PENDING:
client = net.get_ssh(local_ip)
command = 'rm -rf {}'.format(cluster_path)
net.ssh_execute(client=client, command=command)
client.close()
for host in hosts:
if DeployUtil().get_state(cluster_id, host) == PENDING:
client = net.get_ssh(host)
command = 'rm -rf {}'.format(cluster_path)
net.ssh_execute(client=client, command=command)
client.close()
# added_hosts = post_hosts - pre_hosts
msg = message.get('check_cluster_exist')
logger.info(msg)
added_hosts = set(hosts)
meta = []
if deploy_state == DEPLOYED:
pre_hosts = config.get_props(props_path, 'sr2_redis_master_hosts')
added_hosts -= set(pre_hosts)
can_deploy = True
if no_localhost:
added_hosts |= set([local_ip])
for host in added_hosts:
client = net.get_ssh(host)
is_localhost = Center().is_localhost(host)
if is_localhost:
if no_localhost:
continue
if os.path.exists(cluster_path + '/remote'):
meta.append([host, color.green('CLEAN')])
continue
if net.is_exist(client, cluster_path):
meta.append([host, color.red('CLUSTER EXIST')])
can_deploy = False
continue
meta.append([host, color.green('CLEAN')])
if meta:
utils.print_table([['HOST', 'STATUS']] + meta)
if not can_deploy:
msg = message.get('error_cluster_collision')
logger.error(msg)
return
# if not force:
# logger.error("If you trying to force, use option '--force'")
# return
logger.info('OK')
# cluster stop and clean
if deploy_state == DEPLOYED and clean:
center = Center()
cur_cluster_id = config.get_cur_cluster_id(allow_empty_id=True)
run_cluster_use(cluster_id)
center.update_ip_port()
center.stop_redis()
center.remove_all_of_redis_log_force()
center.cluster_clean()
run_cluster_use(cur_cluster_id)
# backup conf
if deploy_state == DEPLOYED:
Center().conf_backup(local_ip, cluster_id, conf_backup_dir)
# backup cluster
backup_hosts = []
if deploy_state == DEPLOYED:
backup_hosts += set(pre_hosts)
# if force:
# backup_hosts += added_hosts
for host in backup_hosts:
cluster_path = path_of_fb['cluster_path']
client = net.get_ssh(host)
Center().cluster_backup(host, cluster_id, cluster_backup_dir)
client.close()
# transfer & install
msg = message.get('transfer_and_execute_installer')
logger.info(msg)
target_hosts = hosts + [local_ip] if no_localhost else hosts
for host in target_hosts:
if not (no_localhost and Center().is_localhost(host)):
logger.info(' - {}'.format(host))
client = net.get_ssh(host)
cmd = 'mkdir -p {0} && touch {0}/.deploy.state'.format(cluster_path)
net.ssh_execute(client=client, command=cmd)
client.close()
DeployUtil().transfer_installer(host, cluster_id, installer_path)
try:
DeployUtil().install(host, cluster_id, installer_name)
except SSHCommandError as ex:
msg = message.get('error_execute_installer')
msg = msg.format(installer=installer_path)
logger.error(msg)
logger.exception(ex)
return
# setup props
if deploy_state == DEPLOYED:
if restore_yes:
tag = conf_backup_dir
else:
tag = tmp_backup_dir
Center().conf_restore(local_ip, cluster_id, tag)
else:
key = 'sr2_redis_master_hosts'
config.make_key_enable(props_path, key)
config.set_props(props_path, key, props_dict['hosts'])
key = 'sr2_redis_master_ports'
config.make_key_enable(props_path, key)
value = cluster_util.convert_list_2_seq(props_dict['master_ports'])
config.set_props(props_path, key, value)
key = 'sr2_redis_slave_hosts'
config.make_key_enable(props_path, key)
config.set_props(props_path, key, props_dict['hosts'])
config.make_key_disable(props_path, key)
if props_dict['replicas'] > 0:
key = 'sr2_redis_slave_hosts'
config.make_key_enable(props_path, key)
key = 'sr2_redis_slave_ports'
config.make_key_enable(props_path, key)
value = cluster_util.convert_list_2_seq(props_dict['slave_ports'])
config.set_props(props_path, key, value)
key = 'ssd_count'
config.make_key_enable(props_path, key)
config.set_props(props_path, key, props_dict['ssd_count'])
key = 'sr2_redis_data'
config.make_key_enable(props_path, key, v1_flg=True)
config.make_key_enable(props_path, key, v1_flg=True)
config.make_key_disable(props_path, key)
config.set_props(props_path, key, props_dict['prefix_of_db_path'])
key = 'sr2_redis_db_path'
config.make_key_enable(props_path, key, v1_flg=True)
config.make_key_enable(props_path, key, v1_flg=True)
config.make_key_disable(props_path, key)
config.set_props(props_path, key, props_dict['prefix_of_db_path'])
key = 'sr2_flash_db_path'
config.make_key_enable(props_path, key, v1_flg=True)
config.make_key_enable(props_path, key, v1_flg=True)
config.make_key_disable(props_path, key)
config.set_props(props_path, key, props_dict['prefix_of_db_path'])
# synk props
msg = message.get('sync_conf')
logger.info(msg)
for node in hosts:
if socket.gethostbyname(node) in config.get_local_ip_list():
continue
client = net.get_ssh(node)
if not client:
msg = message.get('error_ssh_connection').format(host=node)
logger.error(msg)
return
net.copy_dir_to_remote(client, conf_path, conf_path)
client.close()
# set deploy state complete
if os.path.exists(tmp_backup_path):
shutil.rmtree(tmp_backup_path)
for node in target_hosts:
path_of_fb = config.get_path_of_fb(cluster_id)
cluster_path = path_of_fb['cluster_path']
client = net.get_ssh(node)
cmd = 'rm -rf {}'.format(os.path.join(cluster_path, '.deploy.state'))
net.ssh_execute(client=client, command=cmd)
client.close()
if no_localhost:
os.system('touch {}/remote'.format(cluster_path))
msg = message.get('complete_deploy').format(cluster_id=cluster_id)
logger.info(msg)
Cluster().use(cluster_id)
msg = message.get('suggest_after_deploy')
logger.info(msg)
def _deploy_zero_downtime(cluster_id):
logger.debug("zero downtime update cluster {}".format(cluster_id))
center = Center()
center.update_ip_port()
m_hosts = center.master_host_list
m_ports = center.master_port_list
s_hosts = center.slave_host_list
s_ports = center.slave_port_list
path_of_fb = config.get_path_of_fb(cluster_id)
cluster_path = path_of_fb['cluster_path']
# check master alive
m_count = len(m_hosts) * len(m_ports)
alive_m_count = center.get_alive_master_redis_count()
if alive_m_count < m_count:
logger.error(message.get('error_exist_disconnected_master'))
return
if not config.is_slave_enabled:
logger.error(message.get('error_need_to_slave'))
return
# select installer
installer_path = ask_util.installer()
installer_name = os.path.basename(installer_path)
# backup info
current_time = time.strftime("%Y%m%d%H%M%S", time.gmtime())
conf_backup_dir = 'cluster_{}_conf_bak_{}'.format(cluster_id, current_time)
cluster_backup_dir = 'cluster_{}_bak_{}'.format(cluster_id, current_time)
local_ip = config.get_local_ip()
# backup conf
center.conf_backup(local_ip, cluster_id, conf_backup_dir)
# backup cluster
for host in s_hosts:
client = net.get_ssh(host)
center.cluster_backup(host, cluster_id, cluster_backup_dir)
client.close()
# transfer & install
logger.info(message.get('transfer_and_execute_installer'))
for host in m_hosts:
logger.info(' - {}'.format(host))
client = net.get_ssh(host)
cmd = 'mkdir -p {0} && touch {0}/.deploy.state'.format(cluster_path)
net.ssh_execute(client=client, command=cmd)
client.close()
DeployUtil().transfer_installer(host, cluster_id, installer_path)
try:
DeployUtil().install(host, cluster_id, installer_name)
except SSHCommandError as ex:
msg = message.get('error_execute_installer')
msg = msg.format(installer=installer_path)
logger.error(msg)
logger.exception(ex)
return
# restore conf
center.conf_restore(local_ip, cluster_id, conf_backup_dir)
# set deploy state complete
for node in m_hosts:
path_of_fb = config.get_path_of_fb(cluster_id)
cluster_path = path_of_fb['cluster_path']
client = net.get_ssh(node)
cmd = 'rm -rf {}'.format(os.path.join(cluster_path, '.deploy.state'))
net.ssh_execute(client=client, command=cmd)
client.close()
# restart slave
center.stop_current_nodes(master=False, slave=True)
center.configure_redis()
center.sync_conf()
center.start_current_nodes(master=False, slave=True)
center.wait_until_all_redis_process_up()
slaves_for_failover = center.get_slave_nodes()
key = 'cluster-node-timeout'
origin_m_value = center.cli_config_get(key, m_hosts[0], m_ports[0])
origin_s_value = center.cli_config_get(key, s_hosts[0], s_ports[0])
logger.debug('config set: cluster-node-timeout 2000')
RedisCliConfig().set(key, '2000', all=True)
# cluster failover (with no option)
logger.info(message.get('failover_on_deploy'))
logger.debug(slaves_for_failover)
try_count = 0
while try_count < 10:
try_count += 1
success = True
for slave_addr in slaves_for_failover:
host, port = slave_addr.split(':')
stdout = center.run_failover("{}:{}".format(host, port))
logger.debug("failover {}:{} {}".format(host, port, stdout))
if stdout != "ERR You should send CLUSTER FAILOVER to a slave":
# In some cases, the cluster failover is not complete
# even if stdout is OK
# If redis changed to master completely,
# return 'ERR You should send CLUSTER FAILOVER to a slave'
success = False
if success:
break
msg = message.get('retry').format(try_count=try_count)
logger.info(msg)
time.sleep(5)
logger.debug('restore config: cluster-node-timeout')
center.cli_config_set_all(key, origin_m_value, m_hosts, m_ports)
center.cli_config_set_all(key, origin_s_value, s_hosts, s_ports)
if not success:
logger.error(message.get('error_redis_failover'))
return
# restart master (current slave)
center.stop_current_nodes(master=False, slave=True)
center.configure_redis(slave=False)
center.sync_conf()
center.start_current_nodes(master=False, slave=True)
center.wait_until_all_redis_process_up()
def installer():
'''
Select installer from list of '$FBPATH/releases'
or input absolute path of file directly
return installer path
'''
logger.debug('ask installer')
path_of_cli = config.get_path_of_cli(None)
release_path = path_of_cli['release_path']
if not os.path.exists(release_path):
os.mkdir(release_path)
installer_list = net.get_installers_from_fb_s3()
buf = os.listdir(release_path)
buf = list(filter(lambda x: x != '.gitignore', buf))
pattern = '.download'
buf = list(filter(lambda x: pattern not in x, buf))
for file_name in buf:
installer_list.append({
'name': file_name,
'url': os.path.join(release_path, file_name),
'type': 'local',
})
# formatting msg
formatted = []
for i, obj in enumerate(installer_list):
formatted.append(' ({index}) [{type}] {name}'.format(
index=i + 1,
name=obj['name'],
type=obj['type'].upper(),
))
if not formatted:
formatted.append(' (empty)')
stringfied_list = '\n'.join(formatted)
msg = '\n'.join(message.get('ask_installer')).format(list=stringfied_list)
result = ask(msg)
while True:
result = result.strip()
if installer_list and utils.is_number(result):
# case: select in list
result = int(result) - 1
if result in range(0, len(installer_list)):
selected = installer_list[result]
if selected['type'] == 'download':
url = selected['url']
file_name = selected['name']
installer_path = os.path.join(release_path, file_name)
success = net.download_file(url, installer_path)
if success:
logger.info('OK, {}'.format(file_name))
return installer_path
msg = message.get('error_download_installer')
msg = msg.format(url=url)
logger.error(msg)
if selected['type'] == 'local':
ret = selected['url']
logger.debug('Select insaller in list: {}'.format(ret))
logger.info('OK, {}'.format(selected['name']))
return os.path.expanduser(ret)
msg = message.get('error_select_number')
msg = msg.format(max_number=len(installer_list))
logger.error(msg)
elif result.startswith(('~', '/')):
# case: type path
if os.path.isfile(os.path.expanduser(result)):
logger.debug('Select insaller by path: {}'.format(result))
logger.info('OK, {}'.format(os.path.basename(result)))
return os.path.expanduser(result)
msg = message.get('error_type_installer_path')
msg = msg.format(file_path=result)
logger.error(msg)
elif result.startswith(('http://', 'https://')):
# case: type url
url = result
file_name = url.split('?')[0].split('/')[-1]
installer_path = os.path.join(release_path, file_name)
success = net.download_file(url, installer_path)
if success:
logger.info('OK, {}'.format(file_name))
return installer_path
msg = message.get('error_download_installer')
msg = msg.format(url=url)
logger.error(msg)
else:
msg = message.get('error_invalid_input')
msg = msg.format(value=result)
logger.error(msg)
result = ask('')
def __init__(self, target):
self.target = target
logger.error('DuplicatedError: %s' % target)
def slave_ports(cluster_id, m_count, replicas_count):
logger.debug('ask slave ports')
if replicas_count <= 0:
logger.debug('return empty list')
return []
s_count = m_count * replicas_count
start_s_ports = START_PORT + (cluster_id * MASTER_OFFSET) + SLAVE_OFFSET
end_s_ports = start_s_ports + s_count - 1
if start_s_ports == end_s_ports:
default_s_ports = str(start_s_ports)
else:
default_s_ports = '{}-{}'.format(start_s_ports, end_s_ports)
while True:
result = ask(message.get('ask_ports'), default=default_s_ports)
result = list(map(lambda x: x.strip(), result.split(',')))
valid = True
s_ports = set()
p = re.compile('[0-9]+-[0-9]+')
for item in result:
# range number
m = p.match(item)
if m:
s, e = map(int, item.split('-'))
if s > e:
msg = message.get('error_invalid_range').format(value=item)
logger.error(msg)
valid = False
break
s_ports.update(range(s, e + 1))
continue
# single number
elif utils.is_number(item):
s_ports.add(int(item))
continue
else:
msg = message.get('error_invalid_input').format(value=item)
logger.error(msg)
valid = False
break
out_of_range = []
for port in s_ports:
if not port_range_safe(port):
out_of_range.append(port)
if out_of_range:
msg = message.get('error_port_range').format(
minimum=PORT_MININUM,
maximum=PORT_MAXIMUM,
value=out_of_range,
)
logger.error(msg)
continue
if valid and len(s_ports) != s_count:
real_replicas_count = len(s_ports) / float(m_count)
msg = message.get('error_port_count_different').format(
advance=replicas_count,
current=int(real_replicas_count),
)
logger.error(msg)
continue
if valid:
break
s_ports = sorted(list(s_ports))
logger.info('OK, {}'.format(result))
return s_ports
