Nova:
in nova/manager.py
instance usage 发送方式：
method(context, ‘compute.instance.%s‘ % event_suffix, usage_info)
event_suffix = "create.error" / "create.end"
self.notifier = rpc.get_notifier(self.service_name, self.host)
def get_notifier(service, host=None, publisher_id=None):
assert NOTIFIER is not None
if not publisher_id:
publisher_id = "%s.%s" % (service, host or CONF.host)
return NOTIFIER.prepare(publisher_id=publisher_id)
Ceilometer处理:
ceilometer/compute/notification/instance.py
event_types = [‘compute.instance.*‘]
VM privisoning latency:
nova/compute/api.py: def _provision_instances(self, context, instance_type, min_count,
rpc.get_notifier(service, host).info(context,
‘compute.instance.update‘, payload)
是否需要参数：notify_on_state_change
nova/nofitication.py
３中通知
１． notify_on_state_change
(可以是空，vm_state，vm_and_task_state)
用在send_update／send_update_with_states
send_update 用在conductor.manager.py
send_update_with_states在compute.api
发送compute.instance.update类型的消息
最后都是调用
def _send_instance_update_notification中的
rpc.get_notifier(service, host).info(context,
‘compute.instance.update‘, payload)
send_update调用了send_update_with_states，另外send_update_with_states还被用在了nova/compute/api.py 的_provision_instances
notifications.send_update_with_states(context, instance, None,
vm_states.BUILDING, None, None, service="api")
publisherid= service + host
问题是这个在ｃｅｉｌｏｍｅｔｅｒ被转换成ｓａｍｐｌｅ了。
２． notify_api_faults
用在send_api_fault
rpc.get_notifier(‘api‘).error(common_context.get_current() or
nova.context.get_admin_context(),
‘api.fault‘,
payload)
publish ID:‘api‘
类型：‘api.fault‘
用在nova/api/openstack/__init__.py:作为一个FaultWrapper的中间件
etc/nova/api-paste.ini:paste.filter_factory = nova.api.openstack:FaultWrapper.factory
３． default_notification_level
用在def notify_decorator(name, fn):
发送方式publishid=‘api‘+CONF.default_publisher_id or CONF.host 级别info, 类型
nova/utils.py: ‘nova.compute.api:%s‘ % (notify_decorator)
需要配置：
cfg.ListOpt(‘monkey_patch_modules‘,
default=[
‘nova.api.ec2.cloud:%s‘ % (notify_decorator),
‘nova.compute.api:%s‘ % (notify_decorator)
],
4. nova/compute/utils.py
def notify_about_instance_usage
method(context, ‘compute.instance.%s‘ % event_suffix, usage_info)
param event_suffix: Event type like "delete.start" or "exists
被ｗａｐｅｒ
def _notify_about_instance_usage(self, context, instance, event_suffix,
network_info=None, system_metadata=None,
extra_usage_info=None, fault=None):
compute_utils.notify_about_instance_usage(
self.notifier, context, instance, event_suffix,
network_info=network_info,
system_metadata=system_metadata,
extra_usage_info=extra_usage_info, fault=fault)
nova/compute/manager.py:
self._notify_about_instance_usage(context, instance, ‘create.start‘,
create.error
所有的nova.compute.api
需要配置monkey_patch＝true,默认修饰nova.compute.api和nova.api.ec2.cloud
notify_opts = [
cfg.StrOpt(‘notify_on_state_change‘,
help=‘If set, send compute.instance.update notifications on instance ‘
‘state changes. Valid values are None for no notifications, ‘
‘"vm_state" for notifications on VM state changes, or ‘
‘"vm_and_task_state" for notifications on VM and task state ‘
‘changes.‘),
cfg.BoolOpt(‘notify_api_faults‘, default=False,
help=‘If set, send api.fault notifications on caught exceptions ‘
‘in the API service.‘),
cfg.StrOpt(‘default_notification_level‘,
default=‘INFO‘,
help=‘Default notification level for outgoing notifications‘),
cfg.StrOpt(‘default_publisher_id‘,
help=‘Default publisher_id for outgoing notifications‘),
event_type‘: u‘compute.instance.reboot.start
provining vm:
def _build_and_run_instance
- self._notify_about_instance_usage(context, instance, ‘create.start‘,
-- notify_about_instance_usage 发出：
compute.instance.create.start
compute.instance.exists
compute.instance.create.end