原创 Oracle 10g CRS 安装管理一点小结-

shahand

一个网卡拿什么做心跳啊

Yong Huang

最初由 biti_rainy 发布
[B]我怀疑是不是换了个名字，是这个了：

/opt/oracle/products/10.2.0/crs/bin/oclsomon.bin [/B]

You probably meant oclsmon.bin. I don't have that running on my Linux (Oracle 10gR2). I don't use any vendor clusterware.

Do you use vendor clusterware? I wonder if Oracle's statement "this process only appears on platforms that do not use vendor clusterware with Oracle Clusterware" should be the opposite. Note:295326.1 says "The presence of Vendor Clusterware (Serviceguard) should prevent the oprocd from starting".

Yong Huang

shahand

OCR和vd的添加和删除
                shahand  20060425
oracle10gR2开始支持OCR和vd的冗余，voting disk的添加删除比较简单，但ocr我在metalink上没有找到有相关的文章，是摸索过来的，其中还试过用dd来复制ocr的数据，结果crs启动失败了，才根据如何备份ocr想到的ocrconfig命令。

一、ocr的添加。
这里只举例使用raw device的情况
(1)停止crs
＃crsctl stop crs
# rsh lxvcs-8 /OraHome/crs/bin/crsctl stop crs
(2)创建mirror ocr设备，vxassist -g ocrdg make ocrvol1 200M
(3)
使用ocrconfig倒出ocr数据
# ocrconfig -export /oradata/ocr.data
(4)
修改文件，添加第二行
#cat /etc/oracle/ocr.loc
ocrconfig_loc=/dev/vx/rdsk/ocrdg/ocrvol
ocrmirrorconfig_loc=/dev/vx/rdsk/ocrdg/ocrvol1
local_only=FALSE
(5)使用ocrconfig向设备中导入数据
#ocrconfig -import /oradata/ocr.data
(6)此步可选，使用ocrcheck检查
# ocrcheck
Status of Oracle Cluster Registry is as follows :
         Version                  :          2
         Total space (kbytes)     :     204576
         Used space (kbytes)      :       2200
         Available space (kbytes) :     202376
         ID                       : 1831185259
         Device/File Name         : /dev/vx/rdsk/ocrdg/ocrvol
                                    Device/File integrity check succeeded
         Device/File Name         : /dev/vx/rdsk/ocrdg/ocrvol1
                                    Device/File integrity check succeeded

         Cluster registry integrity check succeeded

(7)启动crs，可以用#crsctl start crs
(8)检查crs状态crsctl check crs

二、删除mirror比较简单，去掉/注释掉 ocrmirrorconfig_loc这行，删除然后就可以了

三、voting disk 的添加，见下面的例子，
使用crsctl来添加，唯一要注意的是需要指定-force 参数
[root@lxvcs-8 ~]# crsctl check crs
CSS appears healthy
CRS appears healthy
EVM appears healthy
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]# crsctl query css votedisk
0.     0    /dev/vx/rdsk/ocrdg/vdvol

located 1 votedisk(s).
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]# crsctl add css votedisk /dev/vx/rdsk/ocrdg/vdvol1
Cluster is not in a ready state for online disk addition
[root@lxvcs-8 ~]# crsctl add css votedisk /dev/vx/rdsk/ocrdg/vdvol1 -force
Now formatting voting disk: /dev/vx/rdsk/ocrdg/vdvol1
successful addition of votedisk /dev/vx/rdsk/ocrdg/vdvol1.
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]# crsctl query css votedisk
0.     0    /dev/vx/rdsk/ocrdg/vdvol
1.     0    /dev/vx/rdsk/ocrdg/vdvol1

located 2 votedisk(s).
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]# crsctl check crs
CSS appears healthy
CRS appears healthy
EVM appears healthy

四、voting disk 的删除，同上，同样需要 -force参数
[root@lxvcs-8 ~]# crsctl query css votedisk
0.     0    /dev/vx/rdsk/ocrdg/vdvol
1.     0    /dev/vx/rdsk/ocrdg/vdvol1

located 2 votedisk(s).
[root@lxvcs-8 ~]# crsctl delete css votedisk /dev/vx/rdsk/ocrdg/vdvol1
Cluster is not in a ready state for online disk removal
[root@lxvcs-8 ~]#
[root@lxvcs-8 ~]# crsctl delete css votedisk /dev/vx/rdsk/ocrdg/vdvol1 -force
successful deletion of votedisk /dev/vx/rdsk/ocrdg/vdvol1.

Yong Huang

最初由 Yong Huang 发布
[B]
...I wonder if Oracle's statement "this process only appears on platforms that do not use vendor clusterware with Oracle Clusterware" should be the opposite. Note:295326.1 says "The presence of Vendor Clusterware (Serviceguard) should prevent the oprocd from starting".
[/B]

I think I got the logic wrong. The documentation is consistent with Note:295326.1. One prevents the other from occurring, i.e. either vendor clusterware or oprocd, but not both.

But in my case, I don't see oprocd even though I'm not using vendor clusterware.

Yong Huang

biti_rainy

最初由 Yong Huang 发布
[B]

I think I got the logic wrong. The documentation is consistent with Note:295326.1. One prevents the other from occurring, i.e. either vendor clusterware or oprocd, but not both.

But in my case, I don't see oprocd even though I'm not using vendor clusterware.

Yong Huang [/B]

是的，我在linux上，没有使用第三方 cluster 软件。这个进程是不存在的

在log 里面有如下信息：

[root@forumsb1 root]# cd  /opt/oracle/products/10.2.0/crs/log/forumsb1/cssd/
[root@forumsb1 cssd]# ls -ltr
total 4756
drwxr-xr-x    2 oracle   oinstall     4096 Mar  8 14:19 oclsmon
drwxr-xr-x    2 oracle   root         4096 Mar 21 18:17 oclsomon
-rw-r--r--    1 oracle   oinstall        6 Apr 10 19:07 forumsb1.pid
-rw-r--r--    1 oracle   oinstall       17 Apr 10 19:07 cssdOUT.log
-rw-r--r--    1 oracle   oinstall  2142238 Apr 17 15:24 ocssd.l01
-rw-r--r--    1 oracle   oinstall  2695953 Apr 26 12:35 ocssd.log
[root@forumsb1 cssd]# tail -20  ocssd.log
[    CSSD]2006-04-26 12:22:45.437 [190241712] >TRACE:   clssgmClientConnectMsg: Connect from con(0xb72358c8) proc(0xb7263338) pid() proto(10:2:1:1)
[    CSSD]2006-04-26 12:22:47.912 [190241712] >TRACE:   clssgmClientConnectMsg: Connect from con(0xb72358c8) proc(0xb7263338) pid() proto(10:2:1:1)
[    CSSD]2006-04-26 12:22:57.150 [190241712] >TRACE:   clssgmClientConnectMsg: Connect from con(0xb72358c8) proc(0xb7263338) pid() proto(10:2:1:1)
[    CSSD]2006-04-26 12:23:33.881 [190241712] >TRACE:   clssgmClientConnectMsg: Connect from con(0xb72358c8) proc(0xb7263338) pid() proto(10:2:1:1)

shahand

唉，metalink上文档就是有些细节不太准

cyr1974

1÷这两个文件对于cluster节点是共享的，并且可以同时可读写的吗？
2÷如果不用vendor的ha软件，那么是通过crs来保证cluster节点对该文件的并发读写吗
？

shahand

最初由 cyr1974 发布
[B]1÷这两个文件对于cluster节点是共享的，并且可以同时可读写的吗？
2÷如果不用vendor的ha软件，那么是通过crs来保证cluster节点对该文件的并发读写吗
？ [/B]

1是的
2是由cvm控制的  

askdba

10G R2 CRS 重装一定要FORMAT voting disk ，曾被这个问题折腾了一天，反复检查不知道是什么原因，感觉这点CRS比较傻！

tejun

我在hp安腾主机上安装过10.1，crs的停止用init.crs stop可以停止crs进程，同时也可以停止数据库，但是用init.crs start不能启动crs进程，只能用/etc/inittab中的脚本启动crs，目前在安装10.2在hp安腾平台上，也是init.crs stop可以，但是init.crs start不行，用/etc/inittab中的命令也不行了，尝试了crsctl等命令也不行，现在要把启动crs和数据库的命令都放到双机的脚本中去，却启动不了crs和数据库，郁闷！有什么好的办法吗？