突然掉电导致sybase master库异常求救

xietong007

我们一台服务器上周突然掉电后，sybase就无法启动了，启动日志如下：

00:00000:00000:2014/05/26 18:36:30.57 server  Proc header memory allocated 40632 pages for each per engine cache
00:00000:00000:2014/05/26 18:36:33.40 server  Size of the 8K memory pool: 1625680 Kb
00:00000:00000:2014/05/26 18:36:33.40 server  Memory allocated for the default data cache cachelet 1: 1625680 Kb
00:00000:00000:2014/05/26 18:36:33.40 kernel  Enabling Linux Native Kernel asynchronous disk I/O strategy.
00:00000:00000:2014/05/26 18:36:33.40 kernel  Enabling batch I/O using Linux Native Kernel asynchronous disk I/O strategy.
00:00000:00000:2014/05/26 18:36:33.40 kernel  Initializing virtual device 0, '/zxindata/sybase/data/master.dat' with dsync 'off'.
00:00000:00000:2014/05/26 18:36:33.40 kernel  Virtual device 0 started using asynchronous (with DIRECTIO) i/o.
00:00000:00000:2014/05/26 18:36:33.40 server  Loaded default Unilib conversion handle.
00:00000:00000:2014/05/26 18:36:33.41 kernel  Either the config parameter 'use security services' is set to 0, or ASE does not support use of external security mechanisms on this platform. The Security Control Layer will not be initialized. No external security mechanisms will be supported.
00:00000:00000:2014/05/26 18:36:33.42 kernel  Begin processing to generate RSA keypair.
00:00000:00000:2014/05/26 18:36:33.43 kernel  Unix interval timer enabled for sysclk interrupts.
00:00000:00000:2014/05/26 18:36:33.46 kernel  Completed processing to generate RSA keypair.
00:00000:00000:2014/05/26 18:36:33.46 kernel  Encryption provider initialization succeeded on engine 0.
00:00000:00000:2014/05/26 18:36:33.46 kernel  engine 0, os pid 2229  online
00:00000:00000:2014/05/26 18:36:33.46 server  No active traceflags
00:00000:00001:2014/05/26 18:36:33.47 kernel  libomni1 - Component Integration Services: using 'Sybase Client-Library/15.0/P-EBF16721 ESD #18/DRV.15.0.7/Linux x86_64/Linux 2.4.21-27.ELsmp x86_64 Native Threads/BUILD1500-116/OPT/Sun Mar 22 23:34:35 2009'
00:00000:00001:2014/05/26 18:36:33.47 server  Opening Master Database ...
00:00000:00001:2014/05/26 18:36:33.48 server  Loading ASE's default sort order and character set
00:00000:00001:2014/05/26 18:36:33.49 server  Recovering database 'master'.
00:00000:00001:2014/05/26 18:36:33.49 server  Started estimating recovery log boundaries for database 'master'.
00:00000:00001:2014/05/26 18:36:33.49 server  Database 'master', checkpoint=(3017, 19), first=(3017, 19), last=(3017, 37).
00:00000:00001:2014/05/26 18:36:33.49 server  Completed estimating recovery log boundaries for database 'master'.
00:00000:00001:2014/05/26 18:36:33.49 server  Started ANALYSIS pass for database 'master'.
00:00000:00001:2014/05/26 18:36:33.49 server  Completed ANALYSIS pass for database 'master'.
00:00000:00001:2014/05/26 18:36:33.49 server  Log contains all committed transactions until 2014/05/24 00:30:56.23 for database master.
00:00000:00001:2014/05/26 18:36:33.49 server  Started REDO pass for database 'master'. The total number of log records to process is 19.
00:00000:00001:2014/05/26 18:36:33.49 server  Error: 3474, Severity: 21, State: 1
00:00000:00001:2014/05/26 18:36:33.49 server  During redo the page timestamp value is less than old timestamp from log. Page #=241, object id = 30, page timestamp=0000 10000f703. Log old timestamp=0000 10000f713. Log record marker = (3017, 21).
00:00000:00001:2014/05/26 18:36:33.49 server  Cannot recover the master database, exiting.
00:00000:00001:2014/05/26 18:36:33.49 kernel  ueshutdown: exiting


尝试了在RUN_xxx文件中加入-T3607参数后启动，并使用dump tran master with no_log命令，该命令也无效：

1> dump tran master with truncate_only
2> go
Msg 4237, Level 16, State 1:
Line 1:
Cannot dump master database's log as distributed transaction resolution is not
complete yet. Please retry later.

还请各位大侠看看有没有什么办法恢复？

谢谢！

ceozenliu

用-T3607進去後，試一下：dbcc rebuild_log(master,1,1)

xietong007

ceozenliu 发表于 2014-05-26 15:52
用-T3607進去後，試一下：dbcc rebuild_log(master,1,1)

sybase@SCP2:~/ASE-15_0/install> isql -Usa -P'db10$ZTE'
1> dbcc rebuild_log(master,1,1)
2> go
Database 1 has not been opened so rebuild of log is impossible.
DBCC execution completed. If DBCC printed error messages, contact a user with
System Administrator (SA) role.


试了一下这种办法，提示无法重建log，因为数据库没有打开。 还有没有其他什么招呢？ 谢谢！

ceozenliu

如果有備份的master，就重建master，如果沒有的話，那麼就要聯繫sybase了

xietong007

ceozenliu 发表于 2014-05-26 16:48
如果有備份的master，就重建master，如果沒有的話，那麼就要聯繫sybase了

你好！ 目前就是没有master的备份文件。所以看看有没有其他什么办法恢复。

ceozenliu

這個方法有風險，請參考：http://blog.csdn.net/andkylee/article/details/5004536

ceozenliu

可以再試一下這個方法：
以-T3607進去
一、dbcc checkdb (master)
二、dbcc checktable (sysobjects)
三、dbcc checktable (syslogs)
四、dbcc checktable (sysdatabases)
五、shutdown→restart

試一下吧！

hobbylu

-T3607
dbcc rebuild_log(1,1,1)

jp_1129

1. sybase数据库error 3474的解决。
   
2. 3474错误："During redo the page timestamp value is less than old timestamp from log.Page……"  
   
3. 
   
4. isql -Usa -P -Ssybase :
   
5. 1>sp_configure "allow updates",1
   
6. 2>go
   
7. 
   
8. 1>update sysdatabases set status=-32768 where name="xtsum"
   
9. 2>go
   
10. 1>shutdown
    
11. 2>go
    
12. 重新启动数据库服务。
    
13. isql -Usa -P -Ssybase
    
14. 1>dbcc rebuild_log(xtsum,0,0)
    
15. 2>go
    
16. 1>dbcc rebuild_log(xtsum,1,1)
    
17. 2>go
    
18. 1>shutdown
    
19. 2>g
    
20. 重新启动数据库服务器
    
21. isql -Usa -P -Ssybas
    
22. 1>update sysdatabases set status=12 where name="xtsum"
    
23. 2>g
    
24. 1>sp_configure "allow updates",0  
    
25. 2>
    
26. 1>shutdown
    
27. 2>
    
28. 重启数据库后，恢复正常，rebuild log工作完成。
    
29. 
    
30. 返回结果查询
    
31. select 'truncate table ' + name from sysobjects where type = 'U' and name like 'Tbt_Mk%Up'
    
32. 
    
33. 
    
34. SYBASE3414错误处理:
    
35. 解决办法:
    
36. 重建数据库日志,方法如下:
    
37. 　(1)赋予sa用户sybase_ts_role的角色
    
38. isql -Usa -P
    
39. 1>sp_role "grant","sybase_ts_role",sa
    
40. 2>go
    
41. 1>quit
    
42. 
    
43. (2)将数据库置为"bypass recovery"状态
    
44. isql -Usa -P
    
45. 1>sp_configure "allow updates",1
    
46. 2>go
    
47. 1>use master  
    
48. 2>go  
    
49. 1>update sysdatabases set status=-32768
    
50. 2>where name="database_name"
    
51. 3>go
    
52. 1>shutdown with nowait r`;m'H!  
    
53. 2>go
    
54. (3)rebuild数据库日志
    
55. 重启Server
    
56. isql -Usa -P q=7$ |_.  
    
57. 1>use master
    
58. 2>go
    
59. 1>dbcc rebuild_log(database_name,1,1)
    
60. 2>go
    
61. 1>shutdown with nowait
    
62. 2>go
    
63. (4)重启server
    
64. 1>use master
    
65. 2>go
    
66. 1>update sysdatabases set status=0 where name="database_name"
    
67. 2>go
    
68. 1>sp_configure "allow updates",0
    
69. 2>go
    
70. 1>shutdown with nowait
    
71. 2>go
    
72. (5)在重启server之后,如果数据库恢复正常，rebuild log工作将会成功完成，否则要恢复数据库备份，使用dump database或bcp命令。

复制代码

参见http://blog.chinaunix.net/uid-7509663-id-25471.html

andkylee

dbcc rebuild_log(1,1,1)不好使的话， 修改页面吧。