这样，就不会出现上次rac不能自动启动的问题了。顺便说一下，最后mpath*的权限没有生效，看样子还有点小问题
the udev facility dynamically creates device files in red hat 5. manual changes to the permission, user
and group ids to a device file under /dev are lost after a reboot. to prevent this, create the
/etc/udev/rules.d/99-oracle-permissions.rules file to change the permission, user and group ids. perform
the following steps to change the default permission of the devices on all oracle rac nodes:
通过在目录/etc/udev/rules.d/下添加文件修改磁盘的权限
[root@ rules.d]# vi 99-grid-permissions.rules
# set permissions of block bindings to oracle clusterware devices
program=/bin/chown grid:asmadmin /dev/mapper/mpath1p1
program=/bin/chown grid:asmadmin /dev/mapper/mpath2p1
program=/bin/chown grid:asmadmin /dev/mapper/mpath3p1
program=/bin/chown grid:asmadmin /dev/mapper/mpath4p1
program=/bin/chown grid:asmadmin /dev/mapper/mpath*
~
这样，就不会出现上次rac不能自动启动的问题了。顺便说一下，最后mpath*的权限没有生效，，看样子还有点小问题
参考文献
《deploying oracle rac 10g and oracle rac 11g with asm on red hat enterprise linux using ibm system storage ds8000.pdf》