互联网服务Oracle 系统维护 Oracle rac Oracle集群

oracle RAC报错，给ORACLE迷们解解馋（悬赏100A币）

双机RAC环境
crs运行情况：
bash-3.00# crs_stat -t
Name          Type          Target State    Host
------------------------------------------------------------
ora....B1.lsnr application ONLINE OFFLINE idcdb1 ***
ora.idcdb1.gsd application ONLINE ONLINE idcdb1
ora.idcdb1.ons application ONLINE ONLINE idcdb1
ora.idcdb1.vip application ONLINE ONLINE idcdb2 ***
ora....B2.lsnr application ONLINE ONLINE idcdb2
ora.idcdb2.gsd application ONLINE ONLINE idcdb2
ora.idcdb2.ons application ONLINE ONLINE idcdb2
ora.idcdb2.vip application ONLINE ONLINE idcdb2
ora.orcl.db application ONLINE ONLINE idcdb1
ora....l1.inst application ONLINE ONLINE idcdb1
ora....l2.inst application ONLINE ONLINE idcdb2
bash-3.00#

通过检查idcdb1的日志/home/oracle/app/oracle/product/10.2.0/crs/log/idcdb1/racg/ora.idcdb1.vip.log
当时系统错误日志如下：
2009-01-01 03:16:19.461: [ RACG][1] [8004][1][ora.idcdb1.vip]: ifconfig: SIOCSLIFNAME for ip: fjgi0: already exists
Created new logical interface fjgi0:1
2009-01-19 11:09:52.111: [ RACG][1] [15513][1][ora.idcdb1.vip]: Interface fjgi0 checked failed (host=idcdb1)  ***
Failed to start VIP 192.168.100.34 (host=idcdb1) ***
2009-01-19 11:09:52.117: [ RACG][1] [15513][1][ora.idcdb1.vip]: clsrcexecut: env ORACLE_CONFIG_HOME=/export/home/oracle/app/oracl
e/product/10.2.0/crs
2009-01-19 11:09:52.117: [ RACG][1] [15513][1][ora.idcdb1.vip]: clsrcexecut: cmd = /export/home/oracle/app/oracle/product/10.2.0/
crs/bin/racgeut -e _USR_ORA_DEBUG=0 54 /export/home/oracle/app/oracle/product/10.2.0/crs/bin/racgvip check idcdb1
2009-01-19 11:09:52.117: [ RACG][1] [15513][1][ora.idcdb1.vip]: clsrcexecut: rc = 1, time = 5.975s
2009-01-19 11:09:52.117: [ RACG][1] [15513][1][ora.idcdb1.vip]: end for resource = ora.idcdb1.vip, action = check, status = 1, ti
me = 6.054s
2009-01-19 11:40:17.191: [ RACG][1] [13007][1][ora.idcdb1.vip]: ifconfig: SIOCSLIFNAME for ip: fjgi0: already exists
Created new logical interface fjgi0:1

通过分析：说明在11：09：52时刻，idcdb1上的fjgi0网卡，oracle 检测该网卡时失败。
11：40：17，我通过手动命令方式在idcdb1启动vip

通过检查idcdb2的日志/home/oracle/app/oracle/product/10.2.0/crs/log/idcdb2/racg/ora.idcdb1.vip.log

2008-10-29 15:44:04.731: [ RACG][1] [16813][1][ora.idcdb1.vip]: ifconfig: SIOCSLIFNAME for ip: fjgi0: already exists
Created new logical interface fjgi0:1
2009-01-01 02:05:41.841: [ RACG][1] [12535][1][ora.idcdb1.vip]: ifconfig: SIOCSLIFNAME for ip: fjgi0: already exists
Created new logical interface fjgi0:1
2009-01-19 11:09:36.361: [ RACG][1] [19636][1][ora.idcdb1.vip]: ifconfig: SIOCSLIFNAME for ip: fjgi0: already exists
Created new logical interface fjgi0:2

通过以上日志，可以发现，在11：09：36时刻，oracle在idcdb2上启动了idcdb1上的vip地址，即此刻发生了一次vip地址切换。

分析：通过检查日志情况，可以发现，当oracle check网卡时，若oracle自己认为网卡失效，即是此刻操作系统上网卡未报错，oracle也会认为网卡出错，随即发生vip切换，同时，在失效网卡机器上的监听程序也会失效，从而导致一台机器无法对外提供服务。

谁帮忙分析一下是什么原因造成的，怎么解决。

4 设备清单4.1硬件设备列表

序号	用途	型号	配置	数量
1	数据库服务器	Fujitsu PP650	82.16GHz CPU,16GB内存, 273GB硬盘,两块千兆网卡,2块2GB HBA卡	1
2	数据库服务器	Fujitsu PP650	82.16GHz CPU,12GB内存, 273GB硬盘,两块千兆网卡,2块2GB HBA卡	1
3	光纤交换机	博科BR-200E	82.16GHz CPU,12GB内存, 273GB硬盘,两块千兆网卡,2块2GB HBA卡	1

4.2 系统软件

序号	型号	配置	数量
1	卷管理软件--Symantec SF for oracle rac
2	文件共享软件--Symantec SF CLUSTER FILE SYSTEM

5 主机名和IP划分5.1主机IP地址分配表:

*主机名[/i]***	[i]Public IP	*Private IP*	*VIP*
IDCDB1	192.168.100.34	172.158.100.7	192.168.100.7
IDCDB2	192.168.100.35	172.158.100.6	192.168.100.6

6 主机设备的安装设置6.1
IDCDB1

*服务器用途*	ORACLE RAC Server	*服务器名称*	IDCDB1
*操作系统版本*	Solaris: Version 10.2.0.3.0 SPARC64
*数据库版本*	Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - 64bit
*数据库实例名*	ocrl
IP地址	192.168.100.7	*NetMask*
*Gateway*		*DNS*
*添加路由*
*系统分区*
*编译器*
*JDK*

6.2
IDCDB2

*服务器用途*	ORACLE RAC Server	*服务器名称*	IDCDB2
*操作系统版本*	Solaris: Version 10.2.0.3.0 SPARC64
*数据库版本*	Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - 64bit
*数据库实例名*	ocrl
IP地址	192.168.100.6	*NetMask*
*Gateway*		*DNS*
*添加路由*
*系统分区*
*编译器*
*JDK*