openstack e版创建instance整个流程

感谢朋友支持本博客，欢迎共同探讨交流，由于能力和时间有限，错误之处在所难免，欢迎指正！

如有转载，请保留源作者博客信息。

Better Me的博客：blog.csdn.net/tantexian

如需交流，欢迎大家博客留言。

1、horizon模块：

里面的tables.py的lanch通过url映射到forms.py里面LaunchForm、然后页面传参数进去。然后调用api模块里面nova.py的server_create、

2、python-novaclient

再调用python-novaclient模块V1.1/servers.py的ServerManager类里面的create()、在调用base.py里面的_boot()封装url请求。

3、nova模块：

wsgi对于的发布模块在nova/api/openstack/compute/__init__.py文件里面。发布之后就url地址对应应关系如下：

self._list()对应底层的index()函数     

self._create()对应底层的create()函数

self._delete()对应底层的delete()函数

self._get()对应底层的show()函数

create的url对应nova/api/openstack/compute/servers.py里面的create()

在此函数里面获取image、network、flavor等参数、然后调用self.compute_api.create然后调用self._schedule_run_instance、消息队列发送rpc_method的run_instance方法、根据FLAGS.scheduler_topic配置调用scheduler模块中manager.py的run_instance。配置文件scheduler_driver = FLAGS.scheduler_driver，此处生产环境中用的是--scheduler_driver=nova.scheduler.simple.SimpleScheduler，所以：simple.py里面的schedule_run_instance方法。

driver.cast_to_compute_host将消息发到消息队列中。

compute/manage.py的run_instance方法、_run_instance、_spawn然后配置文件compute_driver = FLAGS.compute_driver的default='nova.virt.connection.get_connection',

get_connection里面 t == 'libvirt':  conn = libvirt_conn.get_connection(read_only) 接着from nova.virt.libvirt import connection as libvirt_conn

因此调用libvirt里面的spawn函数、然后生成xml文件。

1、horizon模块调用流程

2、novaclient模块调用流程

注：        resp, body = self.api.client.post(url, body=body)这句话就是发送post请求，到nova，nova里面的wsgi能够根据这个url自动匹配执行相应函数

3、nova模块调用流程

注:nova的wsgi的发布都在/nova/api/openstack/compute文件夹下面。

上面没有传递available_zone参数????

下面语句则表示调用scheduler

此处生产环境中用的是--scheduler_driver=nova.scheduler.simple.SimpleScheduler，所以：

一下为simple调度算法：

    def _schedule_instance(self, context, instance_opts, *_args, **_kwargs):

        """Picks a host that is up and has the fewest running instances."""

        elevated = context.elevated()

        availability_zone = instance_opts.get('availability_zone')

        

        zone, host = FLAGS.default_schedule_zone, None

        if availability_zone:

            zone, _x, host = availability_zone.partition(':')

        if host and context.is_admin:

            service = db.service_get_by_args(elevated, host, 'nova-compute')

            if not utils.service_is_up(service):

                raise exception.WillNotSchedule(host=host)

            return host

        results = db.service_get_all_compute_sorted(elevated)//获取最少的资源利用率的host

        in_isolation = instance_opts['image_ref'] in FLAGS.isolated_images

        check_cores = not in_isolation or not FLAGS.skip_isolated_core_check

        if zone:

            results = [(service, cores) for (service, cores) in results

                       if service['availability_zone'] == zone] //此处的availability_zone为页面传递过来

        

        requested_networks =  _kwargs['requested_networks']

        if requested_networks:

            results = [(service, cores) for (service, cores) in results

                       if service['network_ref'] in [network_uuid for (network_uuid, fixed_ip) in requested_networks]] //此处的 requested_networks 为页面传递过来

        for result in results:

            (service, instance_cores) = result

            if in_isolation and service['host'] not in FLAGS.isolated_hosts:

                # isloated images run on isolated hosts

                continue

            if service['host'] in FLAGS.isolated_hosts and not in_isolation:

                # images that aren't isolated only run on general hosts

                continue

            if (check_cores and

                instance_cores + instance_opts['vcpus'] > FLAGS.max_cores):

                msg = _("Not enough allocatable CPU cores remaining")

                raise exception.NoValidHost(reason=msg)

            if not self._isenough_subnet(elevated, service):

                continue            

            if utils.service_is_up(service) and not service['disabled']:

                return service['host']

        

        msg = _("Is the appropriate service running?")

        raise exception.NoValidHost(reason=msg)