novaclient的api调用流程与开发

2015年07月05日 19:27:17 阅读数：3915

 

另一个地址：

从nova client的入口查看

cat /usr/bin/nova

 #!/usr/bin/python # PBR Generated from 'console_scripts' import fromimport if:
     sys.exit(main())

其中导入了novaclient.shell这个文件中导入了main方法，进入novaclient.shell.py查看

 def try :]))
 except )
 print"ERROR: %s" )
 .cs = client.Client(options.os_compute_api_version, os_username,
 .extensions, service_type=service_type,
                 cacert=cacert, timeout=timeout)

self.cs是从client中创建出的一个Client实例，进入novaclient.client.py查看这个实例的具体方法

 def : ,
 : ,
 : ,
 try except .join(version_map.keys())))
 raise return def return client_class(*args, **kwargs)

用的是v1_1这个版本的api，对应的是novaclient.v1_1.client.py里的Client类

 class     Top-level object to access the OpenStack Compute API.

     Create an instance with your creds::

         >>> client = Client(USERNAME, PASSWORD, PROJECT_ID, AUTH_URL)

     Then call methods on its managers::

         >>> client.servers.list()
         ...
         >>> client.flavors.list()
         ...

     """

注释里讲了怎么使用python命令行调用nova的client

client里给流入的指令分了很多类，以flavors为例，看nova flavor-list这个命令的流程

 self)
 class     Manage :class:`Flavor` resources.
     """
 def, detailed=, is_public=):
         Get a list of all flavors.

         :rtype: list of :class:`Flavor`.
         """


 ifnot ] = is_public
  % urlutils.urlencode(qparams) ifelse<span "  <="" span="">
 <span "  <="" span="">
 if
 return._list( % (detail, query_string), )

self._list进入novaclient.base.py

 class     Managers interact with a particular type of API (servers, flavors, images,
     etc.) and provide CRUD operations for them.
     """
 def, api):
 .api = api
 def, url, response_key, obj_class=, body=):
 if .api.client.post(url, body=body)
 else .api.client.get(url)
 ifis:
 .resource_class


 if try ]
 except pass .completion_cache(, obj_class, mode=):
 .completion_cache(, obj_class, mode=):
 return, res, loaded=)
 forinif res]

novaclient.v1_1.flavors.py里FlavorManager的resource_class = Flavor即class Flavor(base.Resource)

所以最后obj_class为Flavor

调用api的过程：

 if .api.client.post(url, body=body)
 else .api.client.get(url)

 

 通过.api到了nova的api里nova.api.openstack.compute.__init__.py
 ifis or in .resources[] = flavors.create_resource()
 , ,
 .resources[],
 : },
 : })

 

找到nova.api.openstack.flavors.py

 @wsgi def, req):
 ._get_flavors(req)
 return._view_builder.index(req, limited_flavors)

它最后会返回一个存放flavors信息的字典，这些原始数据经过提取和加工，最后在终端被打印出来

nova.api.openstack.compute.views.flavors.py

 def, func, request, flavors):
 ] forin ._get_collection_links(request,
 ._collection_name,
 )
 if ] = flavors_links
 return flavors_dict

添加一个新的client流程：

功能：快速备份虚拟机，三个参数，虚拟机uuid、备份的名字、备份的描述，调用地方和方法如下：

[plain] 

 

 

 

novaclient.shell.py

class OpenStackComputeShell(object) 的get_subcommand_parser 方法里指定了actions_module

转到novaclient.v1_1.shell.py

增加一个新的方法，装饰器里是需要的参数，有顺序，和执行时的参数顺序一致

 @utils'server', help=)
 @utils'displayname' ,
 )
 @utils'description' ,
 ,
 )
 def                         args.description)

这个功能是加在servers部分里的，转到novaclient.v1_1.servers.py

在ServerManager类里添加

 def, server, backup_name, backup_description):
     Backup a server instance quickly.

     :param server: The :class:`Server` (or its ID) to share onto.
     :param backup_name: Name of the backup image
     :param backup_description: The backup description
     """ : backup_name,
 : backup_description}

 return._create( % base.getid(server),
 )

response_key是指返回数据里的key，这里返回的数据是{'id': "*****"}，所以response_key = "id"

因为这个api返回的是一个json字符串，不能通过novaclient.base.py里Manager类里的方法把数据提取出来（它需要字典），于是把return_raw这个参数设置为True

然后就可以在nova的命令行里看到这个新的功能了：nova backup-instance；使用方法：nova backup-instance <server_id> <bak_name> <bak_description>