存储asm instance spfile到asm，听上去很奇怪，启动自身，首先要解决的几个问题
1,asm diskgroup 没mount,如何读spfile?
2,在oracle_home/dbs下没有参数文件，如何知道spfile在哪?
3,不知道spfile，怎么知道的asm_diskstring?
我在一个11203 2nodes linux rac做个实验
所以得出的结论是
OHASD-->OLR----->GPNP profile(DiscoveryString)--->asm_disk物理路径--->扫描所有disk header,得到spfile物理位置--->直接路径读spfile--->启动ASM--->如有更新同步OLR
reference:
http://aychin.wordpress.com
http://www.askmaclean.com/
1,asm diskgroup 没mount,如何读spfile?
2,在oracle_home/dbs下没有参数文件，如何知道spfile在哪?
3,不知道spfile，怎么知道的asm_diskstring?
我在一个11203 2nodes linux rac做个实验
[grid@znode1 ~]$ crsctl check crs
CRS-4638: Oracle High Availability Services is online
CRS-4537: Cluster Ready Services is online
CRS-4529: Cluster Synchronization Services is online
CRS-4533: Event Manager is online
[grid@znode1 ~]$ crs_stat -t
Name Type Target State Host
------------------------------------------------------------
ora.CRSDG.dg ora....up.type ONLINE ONLINE znode1
ora.DBDG.dg ora....up.type ONLINE ONLINE znode1
ora.FLRV.dg ora....up.type ONLINE ONLINE znode1
ora....ER.lsnr ora....er.type ONLINE ONLINE znode1
ora....N1.lsnr ora....er.type ONLINE ONLINE znode1
ora....N2.lsnr ora....er.type ONLINE ONLINE znode2
ora....N3.lsnr ora....er.type ONLINE ONLINE znode2
ora.asm ora.asm.type ONLINE ONLINE znode1
ora.cvu ora.cvu.type ONLINE ONLINE znode2
ora.gsd ora.gsd.type OFFLINE OFFLINE
ora....network ora....rk.type ONLINE ONLINE znode1
ora.oc4j ora.oc4j.type ONLINE ONLINE znode2
ora.ons ora.ons.type ONLINE ONLINE znode1
ora.rac.db ora....se.type ONLINE ONLINE znode1
ora.scan1.vip ora....ip.type ONLINE ONLINE znode1
ora.scan2.vip ora....ip.type ONLINE ONLINE znode2
ora.scan3.vip ora....ip.type ONLINE ONLINE znode2
ora....SM1.asm application ONLINE ONLINE znode1
ora....E1.lsnr application ONLINE ONLINE znode1
ora.znode1.gsd application OFFLINE OFFLINE
ora.znode1.ons application ONLINE ONLINE znode1
ora.znode1.vip ora....t1.type ONLINE ONLINE znode1
ora....SM2.asm application ONLINE ONLINE znode2
ora....E2.lsnr application ONLINE ONLINE znode2
ora.znode2.gsd application OFFLINE OFFLINE
ora.znode2.ons application ONLINE ONLINE znode2
ora.znode2.vip ora....t1.type ONLINE ONLINE znode2
[grid@znode1 ~]$ srvctl stop database -d rac
[grid@znode1 ~]$ crsctl stop resource ora.FLRV.dg
[grid@znode1 ~]$ crsctl stop resource ora.DBDG.dg
[grid@znode1 ~]$ crsctl stop resource ora.CRSDG.dg
[grid@znode1 ~]$ crsctl stop resource ora.asm
[grid@znode1 ~]$ sqlplus / as sysasm
SQL*Plus: Release 11.2.0.3.0 Production on Fri Jul 20 08:58:45 2012
Copyright (c) 1982, 2011, Oracle. All rights reserved.
Connected to an idle instance.
SQL>
用strace命令跟踪一下进程，也可以用truss,dtrace等平台工具。
[session 2]
[root@znode1 ~]# ps -ef|grep oracle
grid 20642 20639 0 08:58 ? 00:00:00 oracle+ASM1 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))
root 20756 20709 0 09:00 pts/2 00:00:00 grep oracle
[root@znode1 ~]# ps -ef|grep 20639
grid 20639 20638 0 08:58 pts/3 00:00:00 sqlplus as sysasm
grid 20946 20639 0 09:03 ? 00:00:00 oracle+ASM1 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))
root 21460 21401 0 09:09 pts/4 00:00:00 grep 20639
[root@znode1 ~]# strace -o asmstartup.tra -p 20642
Process 20642 attached - interrupt to quit
[session1]
SQL> startup nomount
ASM instance started
Total System Global Area 284565504 bytes
Fixed Size 1344456 bytes
Variable Size 258055224 bytes
ASM Cache 25165824 bytes
SQL>
[session2 显示]
[root@znode1 ~]# strace -o asmstartup.tra -p 20642(不用再重新输入)
Process 20642 attached - interrupt to quit
Process 20642 detached
[root@znode1 ~]#
[root@znode1 ~]# vi asmstartup.tra
read(9, "\\0001\\0\\0\\6\\0\\0\\0\\0\\0\\3\\212\\6\\376\\377\\377\\377\\0\\1\\0\\0\\376\\377\\377\\377\\376\\377\\377\\377\\0\\0\\0"..., 8208) = 49
mmap2(NULL, 196608, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb6ab6000
times({tms_utime=0, tms_stime=3, tms_cutime=0, tms_cstime=0}) = 522261802
...省略
uname({sys="Linux", node="znode1", ...}) = 0
gettimeofday({1342746189, 667048}, NULL) = 0
getuid32() = 1100
getppid() = 20639
...省略
connect(6, {sa_family=AF_FILE, path="/var/tmp/.oracle/sOCSSD_LL_znode1_"...}, 110) = 0
recv(6, "0\\0\\0\\0\\1\\0\\1\\1\\1\\0\\1\\0\\1\\0\\0\\0\\0\\0\\0\\0\\1\\0\\0\\0\\254\\305!\\0\\rq\\0\\0"..., 10240, 0) = 48
recv(6, 0x119c6768, 10240, 0) = -1 EAGAIN (Resource temporarily unavailable)
recv(6, 0x119c6768, 10240, 0) = -1 EAGAIN (Resource temporarily unavailable)
recv(6, 0x119c6768, 10240, 0) = -1 EAGAIN (Resource temporarily unavailable)
recv(6, 0x119c6768, 10240, 0) = -1 EAGAIN (Resource temporarily unavailable)
open("/u01/app/11.2.0/grid/auth/css/znode1/A6126639/9dc2315", O_WRONLY|O_CREAT|O_EXCL, 0644) = 13
fchmod(13, 0644) = 0
write(13, "k\\205\\235\\177", 4) = 4
close(13)
TIP:从这段文本可以看到进程去连接OCSSD 验证自身身份，所以这也是为什么在启asm之前要先启css服务，如果连接失败自然ASM启动就会出错，如果成功就可以通过socket file /var/tmp/.oracle/sCRSD_UI_SOCK与OHASD通信，上篇也说过是OHASD负责OLR管理，所以就可以从OLR中得到SPFILE 路径信息和asm_diskstring，从接下来的输出可以看到
open("/etc/oracle/olr.loc", O_RDONLY) = 13
fstat64(13, {st_mode=S_IFREG|0644, st_size=82, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb6c77000
read(13, "olrconfig_loc=/u01/app/11.2.0/gr"..., 4096) = 82
read(13, "", 4096) = 0
close(13) = 0
munmap(0xb6c77000, 4096) = 0
open("/etc/oracle/ocr.loc", O_RDONLY) = 13
fstat64(13, {st_mode=S_IFREG|0644, st_size=38, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb6c77000
read(13, "ocrconfig_loc=+crsdg\
local_only="..., 4096) = 38
read(13, "", 4096) = 0
close(13) = 0
...
stat64("/etc/localtime", {st_mode=S_IFREG|0644, st_size=405, ...}) = 0
stat64("/u01/app/11.2.0/grid/gpnp/znode1/wallets/peer/cwallet.sso", {st_mode=S_IFREG|0700, st_size=6301, ...}) = 0
lstat64("/u01/app/11.2.0/grid/gpnp/znode1/wallets/peer/cwallet.sso", {st_mode=S_IFREG|0700, st_size=6301, ...}) = 0
open("/u01/app/11.2.0/grid/gpnp/znode1/wallets/peer/cwallet.sso", O_RDONLY|O_LARGEFILE) = 18
write(3, "kfspFileNameGet name=+CRSDG/rac/"..., 76) = 76
write(3, "\
", 1) = 1
times({tms_utime=2, tms_stime=19, tms_cutime=0, tms_cstime=0}) = 522261855
write(3, "kgfspb_shallow_discover dscstr=\\""..., 33) = 33
write(3, "\
", 1) = 1
open("/dev/raw", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY) = 21
fcntl64(21, F_SETFD, FD_CLOEXEC) = 0
getdents64(21, /* 8 entries */, 32768) = 192
getdents64(21, /* 0 entries */, 32768) = 0
close(21) = 0
access("/dev/raw/raw4", R_OK|W_OK) = 0
stat64("/dev/raw/raw4", {st_mode=S_IFCHR|0660, st_rdev=makedev(162, 4), ...}) = 0
open("/dev/raw/raw4", O_RDONLY|O_LARGEFILE) = 21
Tip:
aychin said:"As one would expect OHASD reads this information from OLR, the path to it, it gets from /etc/oracle/olr.loc. I want to note that it is true for Oracle Restart mode, in Oracle RAC environment information is stored in GPNP profile and there is GPNPD process that maintains and manages this profile information"
write(3, "kgfdp_query grp CRSDG", 21) = 21
write(3, "\
", 1) = 1
times({tms_utime=2, tms_stime=20, tms_cutime=0, tms_cstime=0}) = 522261860
write(3, "kgfdp_create", 12) = 12
write(3, "\
", 1) = 1
times({tms_utime=2, tms_stime=20, tms_cutime=0, tms_cstime=0}) = 522261861
write(3, "kgfdp_initCall ausize=1048576 bl"..., 61) = 61
write(3, "\
", 1) = 1
times({tms_utime=2, tms_stime=20, tms_cutime=0, tms_cstime=0}) = 522261861
write(3, "kgfdp_doQuery sync=2", 20) = 20
write(3, "\
", 1) = 1
times({tms_utime=2, tms_stime=20, tms_cutime=0, tms_cstime=0}) = 522261861
write(3, "kgfdp_read", 10) = 10
write(3, "\
", 1) = 1
times({tms_utime=2, tms_stime=20, tms_cutime=0, tms_cstime=0}) = 522261861
write(3, "kgfdp_readMeta totPst=6 blksz=40"..., 34) = 34
write(3, "\
", 1) = 1
TIP：
"kgfspb_shallow_discover dscstr=\\"" 对应的是asm_diskstring,扫描ASM DISK的路径，如果是用ASMLIB或dev/raw/或/dev/oracleasm/disks/在linux环境中值为null,然后扫描了每个磁盘的文件头读取元数据,从文件头可以得到磁盘组信息、spfile、voting disk file 位置,分别是,kfdhdb.dskname,kfdhdb.spfile, kfdhdb.spfflg (first block and number of blocks)and kfdhdb.vfstart, kfdhdb.vfend (begin block and end block). 可以用$ORACLE_HOME/bin下的kfed工具读硬盘文件。
[grid@znode1 ~]$ kfed read /dev/raw/raw1|more
kfbh.type: 1 ; 0x002: KFBTYP_DISKHEAD --Indicates that this is ASM disk header
kfdhdb.dsknum: 0 ; 0x024: 0x0000 -- disk number
kfdhdb.grptyp: 2 ; 0x026: KFDGTP_NORMAL -- Indicates mirroring level, in my case it is NORMAL
kfdhdb.hdrsts: 3 ; 0x027: KFDHDR_MEMBER -- number of member
kfdhdb.dskname: CRSDG_0000 ; 0x028: length=10 -- disk name
kfdhdb.grpname: CRSDG ; 0x048: length=5 -- disk group name
kfdhdb.fgname: CRSDG_0000 ; 0x068: length=10 -- To which failure group this disk belongs
kfdhdb.secsize: 512 ; 0x0b8: 0x0200 -- Disk sector size
kfdhdb.blksize: 4096 ; 0x0ba: 0x1000 -- Disk block size
kfdhdb.ausize: 1048576 ; 0x0bc: 0x00100000 -- Disk au size
kfdhdb.mfact: 113792 ; 0x0c0: 0x0001bc80
kfdhdb.dsksize: 964 ; 0x0c4: 0x000003c4 -- disk total size MB
kfdhdb.vfstart: 256 ; 0x0ec: 0x00000100 -- Begin block address of voting disk file
kfdhdb.vfend: 288 ; 0x0f0: 0x00000120 -- End block address of voting disk file
kfdhdb.spfile: 59 ; 0x0f4: 0x0000003b -- Begin block address of spfile
kfdhdb.spfflg: 1 ; 0x0f8: 0x00000001 -- Number of blocks containing spfile
从X$KFFXP也可以查询到这个位置
X$KFFXP contains the physical allocation table for each ASM file, that is it contains the mapping between ASM files (identified by columns NUMBER_KFFXP and COMPOUND_KFFXP) and their location on disk is
Column Name Description
ADDR table address/identifier
INDX row identifier
INST_ID instance number (RAC)
NUMBER_KFFXP ASM file number. Join with v$asm_file and v$asm_alias
COMPOUND_KFFXP File identifier. Join with compound_index in v$asm_file
INCARN_KFFXP File incarnation id. Join with incarnation in v$asm_file
PXN_KFFXP Extent number per file
XNUM_KFFXP Logical extent number per file (mirrored extents have the same value)
GROUP_KFFXP ASM disk group number. Join with v$asm_disk and v$asm_diskgroup
DISK_KFFXP Disk number where the extent is allocated. Join with v$asm_disk
AU_KFFXP Relative position of the allocation unit from the beginning of the disk. The allocation unit size (1 MB) in v$asm_diskgroup
LXN_KFFXP 0,1 used to identify primary/mirror extent, 2 identifies file header allocation unit (hypothesis)
FLAGS_KFFXP N.K.
CHK_KFFXP N.K.
SQL> select NAME "FILE NAME",
2 AU_KFFXP "AU NUMBER",
3 NUMBER_KFFXP "FILE NUMBER",
4 GROUP_Number,
5 DISK_KFFXP "DISK NUMBER"
6 from x$kffxp, v$asm_alias
7 where GROUP_KFFXP = GROUP_NUMBER
8 and NUMBER_KFFXP = FILE_NUMBER
9 and lower(name) ='registry.253.787925627'
10 ;
FILE NAME AU NUMBER FILE NUMBER GROUP_NUMBER DISK NUMBER
---------------------------------------- ---------- ----------- ------------ -----------
REGISTRY.253.787925627 59 253 1 0
REGISTRY.253.787925627 59 253 1 2
SQL> select group_number,path,disk_number from v$asm_disk order by 1,3;
GROUP_NUMBER PATH DISK_NUMBER
------------ -------------------- -----------
1 /dev/raw/raw1 0
1 /dev/raw/raw2 1
1 /dev/raw/raw3 2
2 /dev/raw/raw4 0
2 /dev/raw/raw5 1
3 /dev/raw/raw6 0
6 rows selected.
SQL> select group_number,name from v$asm_diskgroup;
GROUP_NUMBER NAME
------------ ------------------------------
1 CRSDG
2 DBDG
3 FLRV
Tip:
可以看出spfile可以在/dev/raw/raw1、/dev/raw/raw3 的59号AU上
[grid@znode1 ~]$ cat /etc/oracle/olr.loc
olrconfig_loc=/u01/app/11.2.0/grid/cdata/znode1.olr
crs_home=/u01/app/11.2.0/grid
strings /u01/app/11.2.0/grid/cdata/znode1.olr|grep -I asmparameterfile
cat /u01/app/11.2.0/grid/gpnp/profiles/peer/profile.xml|grep -I asmparameterfile
gpnptool get
都可以得到下面的信息
/u01/app/11.2.0/grid/cdata/znode1.olr、OCR 和/u01/app/11.2.0/grid/gpnp/profiles/peer/profile.xml和spfile中是都同步的么？
做个实验alter system set asm_diskstring='/dev/raw/raw*',发现只有u01/app/11.2.0/grid/gpnp/profiles/peer/profile.xml的输出没有修改。
所以得出的结论是
OHASD-->OLR----->GPNP profile(DiscoveryString)--->asm_disk物理路径--->扫描所有disk header,得到spfile物理位置--->直接路径读spfile--->启动ASM--->如有更新同步OLR
reference:
http://aychin.wordpress.com
http://www.askmaclean.com/
「喜欢这篇文章，您的关注和赞赏是给作者最好的鼓励」
关注作者
【版权声明】本文为墨天轮用户原创内容，转载时必须标注文章的来源（墨天轮），文章链接，文章作者等基本信息，否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容，欢迎发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。
