VCS 5.0 下mount的故障

oioilu

我的架构是双机，都是RHEL 5.2,安装了VCS 5.0,共享一个SAS存储。
SAS存储建立LVM。由于不需要同时读写所以是在cms1主机上建立的LVM，通过'vgchange '把信息广播到cms2.
手工以root身份mount /dev/VolGroup01/LogVol01 到2台主机下的/oradata目录都正常，且可以读写文件。

但是用VCS配置mount该资源，且强制online后，VCS报错为faulted状态。但是查看/oradata确实是已经mount上了，而且是VCS mount的。

报错engine_A.log 信息如下：

1. 
   
2. 2009/01/06 15:40:52 VCS INFO V-16-1-50135 User asdf fired command: hares -clear oradata  from 1.8.8.226
   
3. 2009/01/06 15:40:52 VCS INFO V-16-1-10307 Resource oradata (Owner: unknown, Group: oracle) is offline on cms1 (Not initiated by VCS)
   
4. 2009/01/06 15:41:23 VCS INFO V-16-1-50135 User asdf fired command: hares -online oradata  cms1  from 61.8.168.226
   
5. 2009/01/06 15:41:23 VCS NOTICE V-16-1-10233 Clearing Restart attribute for group oracle on all nodes
   
6. 2009/01/06 15:41:23 VCS NOTICE V-16-1-10301 Initiating Online of Resource oradata (Owner: unknown, Group: oracle) on System cms1
   
7. 2009/01/06 15:41:23 VCS NOTICE V-16-10031-5511 (cms1) Mount:oradata:online:Trying force mount...
   
8. 2009/01/06 15:41:23 VCS NOTICE V-16-10031-5516 (cms1) Mount:oradata:online:Running fsck...
   
9. 2009/01/06 15:41:23 VCS WARNING V-16-10031-5521 (cms1) Mount:oradata:online:Could not mount the block device /dev/VolGroup01/LogVol01.
   
10. 2009/01/06 15:41:24 VCS INFO V-16-2-13001 (cms1) Resource(oradata): Output of the completed operation (online)
    
11. mount: /dev/VolGroup01/LogVol01 already mounted or /oradata busy
    
12. mount: according to mtab, /dev/mapper/VolGroup01-LogVol01 is already mounted on /oradata
    
13. fsck 1.39 (29-May-2006)
    
14. e2fsck 1.39 (29-May-2006)
    
15. Warning!  /dev/VolGroup01/LogVol01 is mounted.
    
16. Warning: skipping journal recovery because doing a read-only filesystem check.
    
17. /dev/VolGroup01/LogVol01: clean, 12/26214400 files, 870584/52428800 blocks
    
18. mount: /dev/VolGroup01/LogVol01 already mounted or /oradata busy
    
19. mount: according to mtab, /dev/mapper/VolGroup01-LogVol01 is already mounted on /oradata
    
20. 2009/01/06 15:43:24 VCS ERROR V-16-2-13066 (cms1) Agent is calling clean for resource(oradata) because the resource is not up even after online completed.
    
21. 2009/01/06 15:43:25 VCS INFO V-16-2-13068 (cms1) Resource(oradata) - clean completed successfully.
    
22. 2009/01/06 15:43:25 VCS INFO V-16-2-13071 (cms1) Resource(oradata): reached OnlineRetryLimit(0).
    
23. 2009/01/06 15:43:25 VCS ERROR V-16-1-10303 Resource oradata (Owner: unknown, Group: oracle) is FAULTED (timed out) on sys cms1
    
24. 2009/01/06 15:43:25 VCS NOTICE V-16-1-10446 Group oracle is offline on system cms1
    
25. 2009/01/06 15:43:25 VCS INFO V-16-6-15004 (cms1) hatrigger:Failed to send trigger for resfault; script doesn't exist
    
26. 2009/01/06 15:43:25 VCS INFO V-16-10031-15005 (cms1) triggers:???:nfs_postoffline:(postoffline) Invoked with arguments cms1, oracle
    
27. 2009/01/06 15:43:25 VCS INFO V-16-6-15002 (cms1) hatrigger:hatrigger executed /opt/VRTSvcs/bin/triggers/nfs_postoffline cms1 oracle   successfully
    
28. 2009/01/06 15:43:25 VCS INFO V-16-6-15004 (cms1) hatrigger:Failed to send trigger for postoffline; script doesn't exist
    
29. 
    

复制代码

main.cf文件如下，很简单，只配置了mount的资源

1. 
   
2. include "types.cf"
   
3. 
   
4. cluster cms (
   
5.         UserNames = { admin = www, asdf = www }
   
6.         Administrators = { admin, asdf }
   
7.         )
   
8. 
   
9. system cms1 (
   
10.         )
    
11. 
    
12. system cms2 (
    
13.         )
    
14. 
    
15. group oracle (
    
16.         SystemList = { cms2 = 0, cms1 = 1 }
    
17.         AutoStartList = { cms2, cms1 }
    
18.         )
    
19. 
    
20.         Mount oradata (
    
21.                 Critical = 0
    
22.                 MountPoint = "/oradata"
    
23.                 BlockDevice = "/dev/VolGroup01/LogVol01"
    
24.                 FSType = ext3
    
25.                 FsckOpt = "-n"
    
26.                 )
    
27. 
    
28. 
    
29. 
    
30.         // resource dependency tree
    
31.         //
    
32.         //      group oracle
    
33.         //      {
    
34.         //      Mount oradata
    
35.         //      }
    
36. 
    

复制代码

[ 本帖最后由 oioilu 于 2009-1-6 15:46 编辑 ]

kns1024wh

LZ强制online时的另一个节点的状态

oioilu

由于只有1个Recource，强制online的时候另外一个node是offline的。
截图如下

[ 本帖最后由 oioilu 于 2009-1-7 10:22 编辑 ]

oioilu

有人知道吗？

david5337

你需要使用LVMLogicalVolume资源类型来定义DG和LV信息！！