openstack快照分析

1.openstack快照

1.1.OpenStack快照类别

注意：不要混淆为卷创建快照和为以卷作为后端存储的虚拟机创建快照

虚拟机快照（快照存放于glance）：nova image-create <server> <name>

• cold snapshot:创建snapshot时，需暂停虚拟机。
• live snapshot with no downtime:创建snapshot时，无需暂停虚拟机。（目前未实现，存在的问题见此链接：http://blog.chinaunix.net/uid-29718549-id-4346700.html）

volume backed以外类型的虚拟机，在nova-api中调用API.snapshot接口创建快照（最终实现在createImage接口）

volume backed的虚拟机，在nova-api中调用API.snapshot_volume_backed接口创建虚拟机快照

卷快照（LVM快照存放于本地）: nova volume-snapshot-create <volume-id>

在nova-api中调用API.volume_snapshot_create接口创建虚拟机快照

1.2.openstack创建快照命令

nova image-create

usage: nova image-create [--show] [--poll] <server> <name>

Create a new image by taking a snapshot of a running server.

Positional arguments

<server>

Name or ID of server.

<name>

Name of snapshot.

Optional arguments

--show

Print image info.

--poll

Report the snapshot progress and poll until image creation is complete.

nova volume-snapshot-create

usage: nova volume-snapshot-create [--force <True|False>]
                                   [--display-name <display-name>]
                                   [--display-description <display-description>]
                                   <volume-id>

Add a new snapshot.

Positional arguments

<volume-id>

ID of the volume to snapshot

Optional arguments

--force <True|False>

Optional flag to indicate whether to snapshot a volume even if its attached to a server. (Default=False)

--display-name <display-name>

Optional snapshot name. (Default=None)

--display-description <display-description>

Optional snapshot description. (Default=None)

1.3.流程

cold snapshot 流程:

# Save the state and stop a running guest, then detach pci devices

      $ virsh   managedsave   vm (managedsave流程见此链接：http://blog.chinaunix.net/uid-29718549-id-4772180.html)

# Create a qemu internal snapshot

      $ qemu-img    snapshot   -c   snap1   vm_snapshot

# Extract the internal snapshot, convert it to qcow2 and export it a file, then upload to glance

      $ qemu-img    convert   -f   qcow2   vm   -O   qcow2   vm_snapshot

# Start the guest again

      $ virsh   start   vm

live snapshot 流程:

# Abort any failed/finished block operations:

     $ virsh blockjob vm vda --abort

# Undefine a running domain. (Note: Undefining a running domain does not _kill_ the domain, it just converts it from persistent to transient.)

     $ virsh undefine vm

# create a destination image with the original backing file and matching size of the instance root disk.

     $ qemu-img create -f qcow2 vm_copy --backing_file=backing_file --size=root_disk_size

#Invoke 'virsh blockcopy' (This will take time, depending on the size of disk image vm1):

     $ virsh blockcopy --domain vm vda vm_copy --wait --verbose

#Abort any failed/finished block operations:

    $ virsh blockjob vm vda --abort

#Define the guest again (to make it persistent):

    $ virsh define vm

#From the obtained new copy, convert the QCOW2 with a backing file to a qcow2 image with no backing file, then upload to glance:

    $ qemu-img convert -f qcow2 -O raw vm_copy vm_convert

2.与传统快照的区别

对openstack而言，虚拟机的快照其实就是镜像。这是因为Openstack是采用创建镜像的方式创建快照（qemu-img convert -f qcow2 -O qcow2 <disk_path>  <out_path>），而不是通过正统的virsh或者qemu的live snapshot创建快照函数( virsh  snapshot-create-as --live ....)来实现的。虚拟机快照做完后以镜像形式存于glance。 openstack的快照与传统快照有以下 区别 ：

• libvirt 主流快照实现： 采用virDomainSnapshotCreateXML()函数(CLI为virsh snapshot-create)。新建的快照与虚拟机有关联：若为内置快照，快照信息和虚拟机存在同一个qcow2镜像中；若为外置快照，新建一个qcow2文件，原虚拟机的disk将变为一个read only的模板镜像，新qcow2镜像仅记录与2.模板镜像的差异数据。这种快照包含快照链信息，可保留disk和ram信息，可回滚至快照点。
• openstack快照实现：openstack并未采用virDomainSnapshotCreateXML()来实现快照，而是单纯的对虚拟机镜像做转换和拷贝，生成一个与虚拟机无关联的镜像，最后上传至glance中。这种快照不包含快照链信息，只保留disk信息，无法回滚至快照点，只能采用该快照镜像创建一个新的虚拟机。

限制与缺点：

• 没有快照链信息，不支持revert恢复虚拟机到某一个快照点
• 只对系统盘进行快照，
• 不支持内存快照，不支持同时对虚拟机和磁盘做快照
• 需要用户进行一致性操作

• 不支持含元数据导出，

• 不支持含元数据导入

• 只支持虚拟机全量数据快照（与快照的实现方式有关，因为是通过image进行保存的）
• 过程较长（需要先通过存储快照，然后抽取并上传至glance)
• 快照以Image方式保存，而非以cinder卷方式保存，无法充分利用存储本身能力加快快照的创建和使用
• 当前限制openstack的虚拟机快照只快照root盘，不快照内存/CPU状态以及挂载磁盘。挂载磁盘需要事先卸载磁盘(数据盘），然后进行快照，然后再挂载磁盘
• 没有快照链信息

3.OpenStack快照源代码流程分析

3.1 nova/compute/api.py

1. # NOTE(melwitt): We don't check instance lock for snapshot because lock is
   
2.     # intended to prevent accidental change/delete of instances
   
3.     @wrap_check_policy
   
4.     @check_instance_cell
   
5.     @check_instance_state(vm_state=[vm_states.ACTIVE, vm_states.STOPPED,
   
6.                                     vm_states.PAUSED, vm_states.SUSPENDED])
   
7.     def snapshot(self, context, instance, name, extra_properties=None):
   
8.         """Snapshot the given instance.
   
9. 
   
10.         :param instance: nova.db.sqlalchemy.models.Instance
    
11.         :param name: name of the snapshot
    
12.         :param extra_properties: dict of extra image properties to include
    
13.                                  when creating the image.
    
14.         :returns: A dict containing image metadata
    
15.         """
16.        #_create_image函数中调用glance api创建image entry，为后将snapshot上传为镜像做准备，虽然镜像和snapshot在可以上传到glance作为镜像启动虚拟机，
17.        #但是为了区分二者的不同，glance将镜像和snapshot标记为不同的类型：type=image 和 type=snapshot
    
18.         image_meta = self._create_image(context, instance, name,
    
19.                                         'snapshot',
    
20.                                         extra_properties=extra_properties)
    
21. 
    
22.         # NOTE(comstud): Any changes to this method should also be made
    
23.         # to the snapshot_instance() method in nova/cells/messaging.py
24.         # 将任务状态(task state) 设置为:image_snapshot_pending
    
25.         instance.task_state = task_states.IMAGE_SNAPSHOT_PENDING
    
26.         instance.save(expected_task_state=[None])
    
27.         #通过rpc调用nova/compute/rpcapi.py的snapshot_instance函数
    
28.         self.compute_rpcapi.snapshot_instance(context, instance,
    
29.                                               image_meta['id

3.2 nova/compute/rpcapi.py

1. #梳理下流程： 
2. （1）用户发起create snapshot的请求; 
3. （2）nova-api服务接收到这个请求并进行前期处理，即3.1中代码的处理流程;
4. （3）真正的snapshot操作是需要在nova-compute节点上执行的，所以nova-api需要向nova-compute发送message
5. #由于OpenStack环境中会有很多个nova-compute,所以需要通过server=_compute_host(None, instance)来获取虚拟机所在的host，并向其发送message。
6. def snapshot_instance(self, ctxt, instance, image_id):
   
7.         version = '3.0'
   
8.         cctxt = self.client.prepare(server=_compute_host(None, instance),
   
9.                 version=version)
   
10.         cctxt.cast(ctxt, 'snapshot_instance',
    
11.                    instance=instance,
    
12.                    image_id=image_id)

3.3 nova/virt/libvirt/driver.py

2.  def snapshot(self, context, instance, image_id, update_task_state):
   
3.         """Create snapshot from a running VM instance.
   
4. 
   
5.         This command only works with qemu 0.14+
   
6.         """
   
7.         try:
   
8.             virt_dom = self._get_domain(instance)
   
9.         except exception.InstanceNotFound:
   
10.             raise exception.InstanceNotRunning(instance_id=instance['uuid'])
    
11. 
    
12.         base_image_ref = instance['image_ref']
    
13. 
    
14.         base = compute_utils.get_image_metadata(
    
15.             context, self._image_api, base_image_ref, instance)
    
16. 
    
17.         snapshot = self._image_api.get(context, image_id)
    
18. 
    
19.         disk_path = libvirt_utils.find_disk(virt_dom)
    
20.         source_format = libvirt_utils.get_disk_type(disk_path)
    
21. 
    
22.         image_format = CONF.libvirt.snapshot_image_format or source_format
    
23. 
    
24.         # NOTE(bfilippov): save lvm and rbd as raw
    
25.         if image_format == 'lvm' or image_format == 'rbd':
    
26.             image_format = 'raw'
    
27. 
    
28.         metadata = self._create_snapshot_metadata(base,
    
29.                                                   instance,
    
30.                                                   image_format,
    
31.                                                   snapshot['name'])
    
32. 
    
33.         snapshot_name = uuid.uuid4().hex
    
34. 
    
35.         state = LIBVIRT_POWER_STATE[virt_dom.info()[0]]
    
36. 
    
37.         # NOTE(rmk): Live snapshots require QEMU 1.3 and Libvirt 1.0.0.
    
38.         # These restrictions can be relaxed as other configurations
    
39.         # can be validated.
    
40.         # NOTE(dgenin): Instances with LVM encrypted ephemeral storage require
    
41.         # cold snapshots. Currently, checking for encryption is
    
42.         # redundant because LVM supports only cold snapshots.
    
43.         # It is necessary in case this situation changes in the
    
44.         # future.
45.         #这里需要注意，解释了为啥现在是cold snapshot而不是live snapshot:
46.         # 有人提过live snapshot的bug，社区认为live snapshot目前不稳定，所以默认条件下采用cold snapshot，并且是通过硬编码来实现的
47.         # 看下面这个判断条件，成立的时候将live_snapshot = true,其中MIN_LIBVIRT_LIVESNAPSHOT_VERSION=1.3.0, 其实现在libvirt的最新版本
    
48.         # 才到1.2.11， 所以这个live_snapshot的条件不满足，就变成了cold_snapshot
    
49.         if (self._host.has_min_version(MIN_LIBVIRT_LIVESNAPSHOT_VERSION,
    
50.                                        MIN_QEMU_LIVESNAPSHOT_VERSION,
    
51.                                        REQ_HYPERVISOR_LIVESNAPSHOT)
    
52.              and source_format not in ('lvm', 'rbd')
    
53.              and not CONF.ephemeral_storage_encryption.enabled):
    
54.             live_snapshot = True
    
55.             # Abort is an idempotent operation, so make sure any block
    
56.             # jobs which may have failed are ended. This operation also
    
57.             # confirms the running instance, as opposed to the system as a
    
58.             # whole, has a new enough version of the hypervisor (bug 1193146).
    
59.             try:
    
60.                 virt_dom.blockJobAbort(disk_path, 0)#live第一步
    
61.             except libvirt.libvirtError as ex:
    
62.                 error_code = ex.get_error_code()
    
63.                 if error_code == libvirt.VIR_ERR_CONFIG_UNSUPPORTED:
    
64.                     live_snapshot = False
    
65.                 else:
    
66.                     pass
    
67.         else:
    
68.             live_snapshot = False
    
69. 
    
70.         # NOTE(rmk): We cannot perform live snapshots when a managedSave
    
71.         # file is present, so we will use the cold/legacy method
    
72.         # for instances which are shutdown.
    
73.         if state == power_state.SHUTDOWN:
    
74.             live_snapshot = False
    
75. 
    
76.         # NOTE(dkang): managedSave does not work for LXC
77.         #注意这里，如果live_snashot目前是false,在做snapshot之前先要执行：
78.         #（1）_detach_pci_devices， 卸载虚拟机挂载的pci设备，比如数据盘
79.         #（2）self._detach_sriov_ports， 卸载虚拟机挂载的SRIOV设备，比如支持SRIOV的网卡设备
    
80.         if CONF.libvirt.virt_type != 'lxc' and not live_snapshot:
    
81.             if state == power_state.RUNNING or state == power_state.PAUSED:
    
82.                 self._detach_pci_devices(virt_dom,
    
83.                     pci_manager.get_instance_pci_devs(instance))
    
84.                 self._detach_sriov_ports(instance, virt_dom)
    
85.                 virt_dom.managedSave(0)#cold第一步
    
86.         #判断虚拟机的后端存储是什么，不同的后端存储做snapshot是不同的，本地文件系统的话，默认qcow2
    
87.         snapshot_backend = self.image_backend.snapshot(instance,
    
88.                 disk_path,
    
89.                 image_type=source_format)
    
90. 
    
91.         if live_snapshot:
    
92.             LOG.info(_LI("Beginning live snapshot process"),
    
93.                      instance=instance)
    
94.         else:
    
95.             LOG.info(_LI("Beginning cold snapshot process"),
    
96.                      instance=instance)
    
97.         #更新任务的状态为：image_pending_upload, 大家都知道做完snapshot要上传
    
98.         update_task_state(task_state=task_states.IMAGE_PENDING_UPLOAD)
99.         #目前做快照的过程是：
100.         #（1）先在../data/nova/instance/snapshots目录下生成临时目录，比如nova/instances/snapshots/tmptHr585
101.         #（2）然后将快照生成到这个目录，具体参见snapshot_backend.snapshot_extract(out_path, image_format)这个函数
102.         #（3）生成完成后，通过glance api上传，具体参见 self._image_api.update
103.         snapshot_directory = CONF.libvirt.snapshots_directory
     
104.         fileutils.ensure_tree(snapshot_directory)
     
105.         with utils.tempdir(dir=snapshot_directory) as tmpdir:
     
106.             try:
     
107.                 out_path = os.path.join(tmpdir, snapshot_name)
     
108.                 if live_snapshot:
     
109.                     # NOTE(xqueralt): libvirt needs o+x in the temp directory
     
110.                     os.chmod(tmpdir, 0o701)
     
111.                     self._live_snapshot(context, instance, virt_dom, disk_path,
     
112.                                         out_path, image_format, base)#热迁移
     
113.                 else:
114.                     #这个函数实际执行了一条命令： qemu-img convert -f qcow2 -O qcow2  disk_path out_path,算是生成了快照
     
115.                     snapshot_backend.snapshot_extract(out_path, image_format)
     
116.             finally:
     
117.                 new_dom = None
     
118.                 # NOTE(dkang): because previous managedSave is not called
     
119.                 # for LXC, _create_domain must not be called.
     
120.                 if CONF.libvirt.virt_type != 'lxc' and not live_snapshot:
     
121.                     if state == power_state.RUNNING:
     
122.                         new_dom = self._create_domain(domain=virt_dom)   ##恢复做快照之前虚拟机的状态
     
123.                     elif state == power_state.PAUSED:
     
124.                         new_dom = self._create_domain(domain=virt_dom,
     
125.                                 launch_flags=libvirt.VIR_DOMAIN_START_PAUSED)
     
126.                     if new_dom is not None:
     
127.                         self._attach_pci_devices(new_dom,
     
128.                             pci_manager.get_instance_pci_devs(instance))
     
129.                         self._attach_sriov_ports(context, instance, new_dom)
     
130.                 LOG.info(_LI("Snapshot extracted, beginning image upload"),
     
131.                          instance=instance)
     
132. 
     
133.             # Upload that image to the image service
     
134. 
     
135.             update_task_state(task_state=task_states.IMAGE_UPLOADING,
     
136.                      expected_state=task_states.IMAGE_PENDING_UPLOAD)
     
137.             with libvirt_utils.file_open(out_path) as image_file:       ###将生成的快照上传到glance
     
138.                 self._image_api.update(context,
     
139.                                        image_id,
     
140.                                        metadata,
     
141.                                        image_file)
     
142.                 LOG.info(_LI("Snapshot image upload complete"),
     
143.                          instance=instance)

参考资料：

1. https://www.redhat.com/archives/libvirt-users/2012-April/msg00142.html

2. http://blog.csdn.net/halcyonbaby/article/details/19998155

2. http://itxx.sinaapp.com/blog/content/130

3. http://kashyapc.com/2012/09/14/externaland-live-snapshots-with-libvirt/

4.http://blog.csdn.net/wsfdl/article/details/38522099

5.http://blog.csdn.net/halcyonbaby/article/details/19998155

6.http://blog.chinaunix.net/uid-20940095-id-4752643.html