您的位置: 首页  >  文章  >  openstack创建instance的流程

openstack创建instance的流程

分类: 文章 • 2024-06-22 21:46:28

http://ezzze.iteye.com/blog/1164920

workflow:

openstack创建instance的流程


1.用户向nova-api发送请求

用户发送请求到nova-api,这里有两种：
a.通过openstack api
从 server.py's controller.create():

Python代码openstack创建instance的流程
  1. self.helper.create_instance(req,body,self.compute_api.create)

create_instance_helper.CreateInstanceHelper() 查表获取基本信息
b.通过ec2 api
从cloud.py.run_instances()

统一调 computer.api.create() 将新的数据插回去

Python代码openstack创建instance的流程
  1. self._ask_scheduler_to_create_instance(context,base_options,
  2. instance_type,zone_blob,
  3. availability_zone,injected_files,
  4. admin_password,image,
  5. instance_id=instance_id,
  6. requested_networks=requested_networks)


2.API 将处理好的数据通过MQ 转发给scheduler .(code from Computer.api)
Python代码openstack创建instance的流程
  1. rpc.cast(context,
  2. FLAGS.scheduler_topic,
  3. {"method":"run_instance",
  4. "args":{"topic":FLAGS.compute_topic,
  5. "instance_id":instance_id,
  6. "request_spec":request_spec,
  7. "availability_zone":availability_zone,
  8. "admin_password":admin_password,
  9. "injected_files":injected_files,
  10. "requested_networks":requested_networks}})

3.Scheduler 获取信息并作出决定 哪一个host 可以来run instance.

Python代码openstack创建instance的流程
  1. def__getattr__(self,key):
  2. returnfunctools.partial(self._schedule,key)

Python代码openstack创建instance的流程
  1. def_schedule(self,method,context,topic,*args,**kwargs):
  2. .......
  3. rpc.cast(context,
  4. db.queue_get_for(context,topic,host),
  5. {"method":method,
  6. "args":kwargs})
  7. LOG.debug(_("Castedto%(topic)s%(host)sfor%(method)s")%locals())

4.Computer 从池中获取信息 并让 Networker 去准备一个ip ,让volume 准备卷, 然后初始化相应的信息,例如创建image,映射device,创建domain,
并将domain 放入running pool中 然后就进入等待直到instance的状态变为running.

a. networker 分配ip
Python代码openstack创建instance的流程
  1. network_info=self.network_api.allocate_for_instance(context,
  2. instance,v*n=is_v*n,
  3. requested_networks=requested_networks)


Python代码openstack创建instance的流程
  1. defallocate_floating_ip(self,context):
  2. returnrpc.call(context,
  3. FLAGS.network_topic,
  4. {'method':'allocate_floating_ip',
  5. 'args':{'project_id':context.project_id}})





b 让 volume 准备卷

Python代码openstack创建instance的流程
  1. bd_mapping=self._setup_block_device_mapping(context,instance_id)
  2. defcreate(self,context,size,snapshot_id,name,description,
  3. volume_type=None,metadata=None,availability_zone=None):
  4. rpc.cast(context,
  5. FLAGS.scheduler_topic,
  6. {"method":"create_volume",
  7. "args":{"topic":FLAGS.volume_topic,
  8. "volume_id":volume['id'],
  9. "snapshot_id":snapshot_id}})

c call nova.virt.libvirt.firewall.IptablesFirewallDriver 建立网络规则
这里是重头戏，单独开个贴记录下....
d call libvirt 创建domian 并launch
Python代码openstack创建instance的流程
  1. domain=self._create_new_domain(xml)
  2. def_create_new_domain(self,xml,persistent=True,launch_flags=0):
  3. ifpersistent:
  4. #Tocreateapersistentdomain,firstdefineit,thenlaunchit.
  5. domain=self._conn.defineXML(xml)
  6. domain.createWithFlags(launch_flags)
  7. else:
  8. #createXMLcallcreatesatransientdomain
  9. domain=self._conn.createXML(xml,launch_flags)
  10. returndomain


e call virt.libvirt.connetion.spwan 等待

Python代码openstack创建instance的流程
  1. defspawn(self,context,instance,network_info,
  2. block_device_info=None):
  3. ..........
  4. def_wait_for_boot():
  5. instance_name=instance['name']
  6. try:
  7. state=self.get_info(instance_name)['state']
  8. exceptexception.NotFound:
  9. msg=_("Duringreboot,%sdisappeared.")%instance_name
  10. LOG.error(msg)
  11. raiseutils.LoopingCallDone
  13. ifstate==power_state.RUNNING:
  14. msg=_("Instance%sspawnedsuccessfully.")%instance_name
  15. LOG.info(msg)
  16. raiseutils.LoopingCallDone
  18. timer=utils.LoopingCall(_wait_for_boot)
  19. returntimer.start(interval=0.5,now=True)


5一旦instance的状态改变至running,他就会去通过networker获取网络信息, 这里有几种方式,取决于你的networkManager

相关推荐