KFED ERROR!!! could not initialize the diag context一例

Maclean Liu(刘相兵

今天想用KFED 的使用出现了以下错误：

[grid@vrh2 trace]$ kfed read /dev/asm-disk1
ERROR!!! could not initialize the diag context

查了一下metalink 说是$ORACLE_HOME/log/disg/asmtool 文件权限的问题：

1.    
   
2.        KFED - ERROR!!! could not initialize the diag context [ID 1335596.1]
   
3. Applies to:  
   
4. Oracle Server - Enterprise Edition - Version: 11.2.0.1.0 and later   [Release: 11.2 and later ]
   
5. Information in this document applies to any platform.
   
6. 
   
7. 
   
8. Symptoms
   
9. kfed command shows the following error even though the running user has a proper permission on the target device.
   
10. 
    
11. 
    
12. 
    
13. 
    
14.   $ kfed read /u02/oradata/asm_disk0
    
15. ERROR!!! could not initialize the diag context
    
16. 
    
17. 
    
18. 
    
19.   $ id
    
20. uid=503(grid) gid=501(oinstall) groups=501(oinstall),504(asmadmin),505(asmoper),506(asmdba)
    
21. 
    
22. 
    
23. 
    
24.   $ ls -l /u02/oradata/asm_disk0
    
25. -rwxrwxrwx 1 grid oinstall 2097152000 Jul 5 22:14 /u02/oradata/asm_disk0
    
26. 
    
27. Cause
    
28. kfed executable will need create a directory for its own trace files
    
29. under <GridHome>/log/diag/asmtool. If kfed firstly was used as
    
30. root user, this directory - <GridHome>/log/diag/asmtool is created
    
31. as root.
    
32. The following kfed run by other than root user will not be able to create a directory under <GridHome>/log/diag/asmtool.
    
33. 
    
34. 
    
35. 
    
36.   
    
37. $ ls -l $ORACLE_HOME/log/diag
    
38. 
    
39. drwxrwxrwx 2 grid oinstall 4096 Jul 5 15:29 asm
    
40. drwxr-xr-x
    
41. 3 root root 4096 Jul 6 11:17 asmtool << This directory is owned
    
42. by root and there is no write permissin for other users.
    
43. drwxr-xr-x 5 oracle asmadmin 4096 Jun 20 09:16 rdbms
    
44. drwxrwxrwx 3 grid oinstall 4096 Jun 15 18:04 tnslsnr
    
45. 
    
46. Solution
    
47. Change the owner for <GridHome>/log/diag/asmtool directory as grid user.
    
48. 
    
49. # chown grid:oinstall $ORACLE_HOME/log/disg/asmtool
    
50. 
    
51. And as grid user, the following command will show the correct kfed output.
    
52. 
    
53. 
    
54.   
    
55. $kfed read /u02/oradata/asm_disk0 | more
    
56. kfbh.endian: 1 ; 0x000: 0x01
    
57. kfbh.hard: 130 ; 0x001: 0x82
    
58. kfbh.type: 1 ; 0x002: KFBTYP_DISKHEAD
    
59. kfbh.datfmt: 1 ; 0x003: 0x01
    
60. kfbh.block.blk: 0 ; 0x004: T=0 NUMB=0x0
    
61. kfbh.block.obj: 2147483648 ; 0x008: TYPE=0x8 NUMB=0x0
    
62. kfbh.check: 1175212953 ; 0x00c: 0x460c5399
    
63. kfbh.fcn.base: 0 ; 0x010: 0x00000000
    
64. kfbh.fcn.wrap: 0 ; 0x014: 0x00000000
    
65. kfbh.spare1: 0 ; 0x018: 0x00000000
    
66. kfbh.spare2: 0 ; 0x01c: 0x00000000
    
67. kfdhdb.driver.provstr: ORCLDISK ; 0x000: length=8
    
68. ..

复制代码

查了以下 没有$ORACLE_HOME/log/disg/asmtool  这个目录


grid@vrh2 diag]$ ls -l $ORACLE_HOME/log/disg/asmtool
ls: /g01/11.2.0/grid/log/disg/asmtool: No such file or directory


之后对 kfed 做了 strace 才发现 文档写错了：

[grid@vrh2 diag]$ strace kfed

strace log:

stat("/g01/orabase/diag/asmtool", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
stat("/g01/orabase/diag/asmtool/user_grid", 0x7fffe3fe1250) = -1 ENOENT (No such file or directory)
mkdir("/g01/orabase/diag/asmtool/user_grid", 0775) = -1 EACCES (Permission denied)
fstat(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fb6d1335000
write(1, "ERROR!!! could not initialize th"..., 48ERROR!!! could not initialize the diag context
) = 48
close(3)                                = 0
munmap(0x7fb6d10fd000, 143360)          = 0
exit_group(0)                           = ?


应当是 $ORACLE_BASE/diag/asmtool 这个目录才对

[grid@vrh2 ~]$ ls -l $ORACLE_BASE/diag/asmtool
total 4
drwxr-xr-x 3 root root 4096 Jan  4 18:33 user_root


要给予grid 用户合适的权限


[root@vrh2 ~]# chown grid:oinstall $ORACLE_BASE/diag/asmtool
[root@vrh2 ~]# chown grid:oinstall /g01/orabase/diag/asmtool/user_root

[grid@vrh2 ~]$ kfed
as/mlib         ASM Library [asmlib='lib']
aun/um          AU number to examine or update [AUNUM=number]
aus/z           Allocation Unit size in bytes [AUSZ=number]
blkn/um         Block number to examine or update [BLKNUM=number]
blks/z          Metadata block size in bytes [BLKSZ=number]
ch/ksum         Update checksum before each write [CHKSUM=YES/NO]
cn/t            Count of AUs to process [CNT=number]
de/v            ASM device to examine or update [DEV=string]
dm/pall         Don't suppress repeated lines when dumping corrupt blocks [DMPALL=YES/NO]
o/p             KFED operation type [OP=READ/WRITE/MERGE/REPAIR/NEW/FORM/FIND/STRUCT]
p/rovnm         Name for provisioning purposes [PROVNM=string]
s/eek           AU number to seek to [SEEK=number]
te/xt           File name for translated block text [TEXT=string]
ty/pe           ASM metadata block type number [TYPE=number]