def syn_vip(self, clu_name):
"""
获取cluster的vip
:param clu_name:
:return:
"""
rlt = LauncherClient.instance().get_cluster_info(clu_name)
if not rlt.success:
Log(1, "syn_vip get_cluster_info error:{}".format(rlt.message))
return
json_data, s = self.parse_json(rlt.content.get('info'))
if not s:
Log(1, "syn_vip cluster_info can not parse to json:{}".format(rlt.content.get('info')))
return
vip = json_data.get('vip')
rlt = Clusterdb.instance().get_vip(clu_name)
if rlt.success:
vip0, s = self.parse_json(rlt.content)
if not s:
Log(1, "syn_vip the info can not parse to json:{}".format(rlt.content))
return
else:
if rlt.result == ETCD_KEY_NOT_FOUND_ERR:
vip0 = ''
else:
Log(1, "syn_vip can get vip error:{}".format(rlt.message))
return
if vip != vip0:
Clusterdb.instance().save_vip(clu_name, {'vip': vip})
return
def get_api(self, cluster_name):
rlt = LauncherClient.instance().get_cluster_auth_info(cluster_name)
if not rlt.success:
return rlt
# Log(4, "get_cluster_client auth_info:{}".format(rlt.content))
client = KubeClient(rlt.content)
con = client.connect()
if con.success:
return Result(client.api)
else:
Log(
3, 'KubeClientMgr get_cluste_client [%s]fail, as[%s]' %
(cluster_name, con.message))
return Result('', 500, con.message)
def load_cluster(self, cluster_name):
"""
加载单个集群client
:param cluster_name:
:return:
"""
rlt = LauncherClient.instance().get_cluster_auth_info(cluster_name)
if not rlt.success:
if rlt.result == CLU_IS_PENDING:
return Result('', CLU_IS_PENDING, 'clu master is pending')
Log(
1,
'KubeClientMgr.load_cluster read_cluster_auth_info[%s]fail,as[%s]'
% (cluster_name, rlt.message))
return rlt
Log(4, "load_cluster, cluster_name:{}".format(cluster_name))
return self.add_cluster_client(cluster_name, rlt.content)
def timeout(self):
try:
Log(3, "syndata #timeout start at:{}".format(datetime.datetime.now()))
if not CluNodedb.instance().ismaster():
Log(3, "syndata this node is not master")
return
# 当队列中有任务不添加
if self.task_queue.qsize():
Log(3, "syndata timeout task_queue.qsize:{},".format(self.task_queue.qsize()))
return
clu_apply_num = {}
rlt1 = Clusterdb.instance().read_clu_map()
if not rlt1.success:
Log(1, "syndata timeout rlt1.message:{}".format(rlt1.message))
return
for c, v in rlt1.content.items():
sp_key = c.split('/')
if sp_key[-1] == 'apply_num':
clu_apply_num[sp_key[-2]] = v
clu_node = {}
rlt = CluNodedb.instance().read_clunode_map(pass_nll_value=False)
if not rlt.success:
return Result('', rlt.result, rlt.message, 500)
if rlt.success:
for k, v in rlt.content.items():
sp_key = k.split('/')
if len(sp_key) == 6:
clu_node.setdefault(sp_key[4], {})[sp_key[5]] = v
if len(sp_key) == 5:
clu_node.setdefault(sp_key[4], {})
# apply_num个数的参数
clu_ws = WorkSpacedb.instance().read_gws()
if not clu_ws.success:
Log(3, "syndata timeout ws :{}".format(clu_ws.message))
return
for clu_name, nodes in clu_node.items():
# 只同步有主机的集群
if not nodes:
continue
# 更新k8s集群的vip
# self.syn_vip(clu_name)
# 同步更新集群应用个数
syn_clu = self.syn_apply_num(clu_name, clu_ws.content.get(clu_name, []), clu_apply_num.get(clu_name, 0))
if not syn_clu.success:
Log(1, "syndata clu_info apply_num error:{}".format(syn_clu.message))
# 当apiserver连接不上时，表明集群异常(可能是刚添加，也可能是添加成功后出现异常)，需要修改主机的状态
rlt = KubeClientMgr.instance().get_cluster_client(clu_name)
if not rlt.success:
Log(3, "rlt.message:{}".format(rlt.message))
# 如果集群状态是pending则不执行检查任务
if rlt.result == CLU_IS_PENDING:
return
check_num = 3
while check_num:
Log(3, "check_num:{}, clu_name:{}, node:{}".format(check_num, clu_name, len(nodes)))
rlt = KubeClientMgr.instance().get_cluster_client(clu_name)
if rlt.success:
break
check_num -= 1
time.sleep(0.5)
if check_num == 0:
self.change_node_status(nodes, {'status': 'error', 'message': rlt.message, 'pod_num': 0})
continue
client = rlt.content
gws = clu_ws.content.get(clu_name, [])
ws_list = []
for i in gws:
ws_list.extend(i.get('workspace', []))
for n in nodes.values():
if not n['name']:
rlt = LauncherClient.instance().get_host_name(clu_name, n['type'] + 's', n['ip'])
if not rlt.success:
continue
host_name = rlt.content
ip_name = n['ip'].replace('.', '-')
rlt = CluNodedb.instance().update_node(clu_name, ip_name, {'name': host_name})
if not rlt.success:
Log(1, "update_node error:{}".format(rlt.message))
n['name'] = host_name
task = SynDataFac(clu_name, n, client, ws_list)
self.create_task(task)
Log(3, "syndata create task all done at:{}".format(datetime.datetime.now()))
return True
except Exception as e:
Log(1, "sysdata error:{}".format(e.message))
PrintStack()
return None
except KeyboardInterrupt:
Log(3, "syndata timeout be killed")
return None
def create_new_cluster(self, cluster_info, passport):
"""
创建集群
:param creater:
:param cluster_info:
:return:
"""
master_ip = cluster_info.get('addr', '').split(':')[0]
host_name = master_ip.replace('.', '-')
# 检查license
if not passport.get('licensed', ''):
return Result('', LICENSE_OUT_OF_DATE, 'licensed is out of date',
400)
# check集群是否存在
cluster_name = cluster_info.get('cluster_name', '')
if Clusterdb.instance().clu_is_exist(cluster_name):
return Result(0, CLUSTER_HAS_EXISTED, 'clu is existed', 400)
masternode_list = []
nodemonitor_list = []
clunode_list = []
if cluster_info.get('create_way', '') == 'add':
# check 集群ip是否添加过
if CluNodedb.instance().is_node_exist(cluster_name, host_name):
return Result(0, msg='', result=CLUSTER_HAS_EXISTED, code=400)
# 检查是否是ufleet主机
ufleet_hosts = GetSysConfig('ufleet_hosts').split(',')
if master_ip in ufleet_hosts:
return Result('',
msg='the host is used by ufleet.',
result=NODE_USED_BY_UFLEET,
code=400)
client = KubeClient({
'auth_data':
cluster_info.get('cacerts', ''),
'server':
'https://' + cluster_info.get('addr', ''),
'cert_data':
cluster_info.get('apiservercerts'),
'client_key':
cluster_info.get('apiserverkey'),
'cluser_name':
cluster_name
})
rlt = client.connect()
if not rlt.success:
Log(
3, 'KubeClientMgr.add_cluster[%s]fail, as[%s]' %
(cluster_name, rlt.message))
return rlt
self.__store[cluster_name] = client
rlt = client.get_all_nodes()
if not rlt.success:
return rlt
for j in rlt.content:
address = j.get('status', {}).get('addresses', [])
for add in address:
if 'InternalIP' == add.get('type', ''):
ip = add.get('address')
if ip == cluster_info.get('addr', '').split(':')[0]:
host_type = 'master'
else:
host_type = 'node'
ip_name = ip.replace('.', '-')
node_data = node_struct(cluster_name,
add.get('address'), host_type,
cluster_info.get('creater'))
node_data = self.syn_nodeinfo(node_data, j, [])
# clusternode
clunode_list.append({
'cluster_name': cluster_name,
'data': node_data
})
# masternodedir
masternode_data = masternode_struct(
cluster_info.get('creater'), cluster_name,
host_type, add.get('address',
''), '', '', '', '', '', '')
masternode_list.append({
'master_ip': ip_name,
'data': masternode_data
})
# nodemonitor
nodemonitor_list.append(ip_name)
# 调用launcher保存集群认证信息接口
auth_data = auth_info_struct(cluster_info)
rlt = LauncherClient.instance().load_cluster(auth_data)
if not rlt.success:
return Result('', 500, 'load_cluster error:' + rlt.message,
500)
# 保存数据到etcd
new_clu = clu_struct(cluster_info)
rlt = Clusterdb.instance().create_cluster_full(cluster_name, new_clu)
if not rlt.success:
return Result('', rlt.result, rlt.message, 400)
for i in clunode_list:
rlt = CluNodedb.instance().save_node(i['cluster_name'], i['data'])
if not rlt.success:
return rlt
for i in masternode_list:
rlt = Masterdb.instance().save_master(i['master_ip'], i['data'])
if not rlt.success:
return rlt
return Result('')