技术博客 Blog 虚拟机备份

OpenStack计算节点挂掉之后，虚拟机恢复方式

下载免费试用版

适用于多种数据备份

OpenStack计算节点挂掉之后，虚拟机恢复方式

2020-12-09

#环境信息

OS --> CentOS7.2-1511
OpenStack --> Mikata
Ceph -->j版
虽然OpenStack自带有迁移和疏散机制，但并不一定保证100%成功，本文基于疏散失败的情况，来恢复实例。

起因客户那边，物理机系统盘故障，导致数据全部丢失，最开始想到的方法是疏散，直接在dashboard或者控制节点终端执行疏散命令

nova evacuate  node09

然而事情并没有到此结束，有2台虚拟机，一直疏散不了，没办法，数据最重要，由于nova，glance，cinder使用的都是ceph，即数据都存放在ceph中，这样恢复起来就有了可能，
具体步骤如下：

重建libvirt.xml文件
首先查看一下拟机信息

nova show 6bb3bc65-91c3-486c-b853-2a5c879a5395

具体信息如下：

+--------------------------------------+----------------------------------------------------------+
| Property                             | Value                                                    |
+--------------------------------------+----------------------------------------------------------+
|  network                             | 192.168.10.25, xx.xx.xx.xx                              |
| OS-DCF:diskConfig                    | AUTO                                                     |
| OS-EXT-AZ:availability_zone          | nova                                                     |
| OS-EXT-SRV-ATTR:host                 | node03                                                   |
| OS-EXT-SRV-ATTR:hypervisor_hostname  | node03                                                   |
| OS-EXT-SRV-ATTR:instance_name        | instance-000001ae                                        |
| OS-EXT-STS:power_state               | 1                                                        |
| OS-EXT-STS:task_state                | -                                                        |
| OS-EXT-STS:vm_state                  | active                                                   |
| OS-SRV-USG:launched_at               | 2017-06-23T06:03:20.000000                               |
| OS-SRV-USG:terminated_at             | -                                                        |
| accessIPv4                           |                                                          |
| accessIPv6                           |                                                          |
| config_drive                         |                                                          |
| created                              | 2017-06-23T06:01:44Z                                     |
| flavor                               | yunwei01 (d76ea4cd-0c46-423c-9a2a-0ab15c5a1b0a)          |
| hostId                               | 7081812a3c654f417fc545300ecd03252d8ac4bf992b54272bcfee61 |
| id                                   | 6bb3bc65-91c3-486c-b853-2a5c879a5395                     |
| image                                | centos65 (0488b591-1755-4da0-abe9-8e5c2a6931b5)          |
| key_name                             | -                                                        |
| metadata                             | {}                                                       |
| name                                 | rhea-2                                                   |
| os-extended-volumes:volumes_attached | [{"id": "7cd2936a-b4ef-4e85-8159-eace4c1b7981"}]         |
| progress                             | 0                                                        |
| security_groups                      | default                                                  |
| status                               | ACTIVE                                                   |
| tenant_id                            | 290f841df5644709847a51d6604a228f                         |
| updated                              | 2017-12-21T06:46:13Z                                     |
| user_id                              | d158732076c740aead5286969273faea                         |
+-------------------------------------+----------------------------------------------------------+

记录这些信息，后面有大用

在平台上，找一个flavor和image相同的实例，找到此实例uuid
查看该实例信息

nova show uuid

进入到相应节点,拷贝libvirt.xml和console.log

cd /var/lib/nova/instance/$uuid/cp * /mnt/bakrecoverycd /mnt/bakrecoverymore libvirt.xml

信息如下：


  6bb3bc65-91c3-486c-b853-2a5c879a5395
  instance-000001ae
  16777216
  16
  
    
      
      rhea-2
      2017-12-21 06:46:11
      
        16384
        40
        2048
        0
        16
      
      
        admin
        admin
      
      
    
  
  
    
      Fedora Project
      OpenStack Nova
      13.1.0-1.el7
      d3c7a080-2534-4b5f-9d86-9cd152acc671
      6bb3bc65-91c3-486c-b853-2a5c879a5395
      Virtual Machine
    
  
  
    hvm
    
    
  
  
    
    
  
  
    16384
  
  
    
    
    
  
  
    
  
  
    
      
      
        
        
        
      
      
        
      
      
    
    
      
      
        
        
        
      
      
        
      
      
    
    
      
      
        
        
        
      
      
        
      
      
      7cd2936a-b4ef-4e85-8159-eace4c1b7981

需要修改如下部分：
1.name和uuid

  6bb3bc65-91c3-486c-b853-2a5c879a5395
  instance-000001ae

前面记录信息，刚好派上用场
2.name和创建时间

rhea-2
      2017-12-21 06:46:11

这个name是创建实例的时候，我们起的名字，上面的那个，是随机生成的名字
3.修改相应uuid（包括image，project，user）

admin
        admin

3.修改虚拟机uuid，和serial，serial通过比对发现，同一个计算节点这个号相同

     d3c7a080-2534-4b5f-9d86-9cd152acc671
      6bb3bc65-91c3-486c-b853-2a5c879a5395
      Virtual Machine4.块设备id，格式为虚拟机uuid_disk,uuid_swap
5.通过cinder划分的卷id

      7cd2936a-b4ef-4e85-8159-eace4c1b7981
    
6.虚拟机网卡mac地址，还有端口，这里，mac地址可通过dashboard查看，admin账户下，管理员–>网络–>（网络）uuid–>端口

      
      
      
      
7.console日志路径
source path="/var/lib/nova/instances/6bb3bc65-91c3-486c-b853-2a5c879a5395/console.log"/>
8.对比修改完后，一定要仔细确认，以免出现问题
mkdir -p /var/lib/nova/instance/$uuidcp libvirt.xml  /var/lib/nova/instance/$uuidcp console.log /var/lib/nova/instance/$uuid
修改权限
chown nova.nova /var/lib/nova/instance/$uuid/xmlchown qemu.qemu /var/lib/nova/instance/$uuid/console.log
在数据库中修改节点信息
update instances set host='node10',node='node10' where uuid='6bb3bc65-91c3-486c-b853-2a5c879a5395';
重启相应节点nova-coompute服务
systemctl restart openstack-nova-compute
启动实例，启动后网络可能不通，这时可以通过迁移来解决，