双机断开网络后的包切换问题

blacksabbather

大家好！
我配了一个双节热备，上面跑oracle
由于是远程连上去的，所以做网络测试的时候使用交换机端口关闭的方式。
当我把Node A连接到交换机的端口“disable”之后，它会认为Node B “fail”了 Node B会被重启，而Node A虽然网络已断，但是什么都不会做
反过来，如果oracle包在Node B上运行 把Node B的端口disable后Node B会老老实实交出锁盘并重启，Node A会接管
下面是我disable Node A后syslog.log中的关键内容：
Feb 4 15:09:39 Ecards1 cmnetd[3442]: lan1 failed
Feb 4 15:09:39 Ecards1 cmnetd[3442]: Subnet 192.168.180.0 down
Feb 4 15:09:39 Ecards1 cmcld[3434]: Subnet 192.168.180.0 in package oracle is d
own.
Feb 4 15:09:39 Ecards1 cmcld[3434]: Failing package oracle on node Ecards1 due
to subnet failure.
Feb 4 15:09:39 Ecards1 cmcld[3434]: Request from node Ecards1 to fail package o
racle on node Ecards1.
Feb 4 15:09:39 Ecards1 cmcld[3434]: Executing '/etc/cmcluster/oracle/control.sh
stop' for package oracle, as service PKG*42753.
Feb 4 15:09:39 Ecards1 syslog: cmhaltserv ora92in
Feb 4 15:09:39 Ecards1 syslog: cmhaltserv listener
Feb 4 15:09:39 Ecards1 su: + tty?? root-oracle
Feb 4 15:09:40 Ecards1 syslog: cmmodnet -r -i 192.168.180.13 192.168.180.0
Feb 4 15:09:40 Ecards1 cmcld[3434]: Service PKG*42753 terminated due to an exit
(1).
Feb 4 15:09:40 Ecards1 cmcld[3434]: Halted package oracle on node Ecards1.
Feb 4 15:09:39 Ecards1 su: + tty?? root-oracle
Feb 4 15:09:40 Ecards1 cmcld[3434]: Package oracle halt script exited with NO_R
ESTART.
Feb 4 15:09:40 Ecards1 cmcld[3434]: Examine the file /etc/cmcluster/oracle/cont
rol.sh.log for more details.
Feb 4 15:09:40 Ecards1 cmcld[3434]: Switching disabled on package oracle.
Feb 4 15:09:40 Ecards1 cmcld[3434]: Request from node Ecards1 to disable global
Feb 4 15:09:42 Ecards1 cmcld[3434]: Timed out node Ecards2. It may have failed.
Feb 4 15:09:42 Ecards1 cmcld[3434]: Attempting to adjust cluster membership
Feb 4 15:09:42 Ecards1 cmcld[3434]: Beginning standard partial election
Feb 4 15:09:43 Ecards1 cmcld[3434]: Sending file $SGRUN/frdump.cmcld.6 (161466
bytes) to file assistant daemon.
Feb 4 15:09:43 Ecards1 cmfileassistd[4542]: Changed to working directory /var/a
dm/cmcluster/cmfileassistd.
Feb 4 15:09:43 Ecards1 cmfileassistd[4542]: Updated file /var/adm/cmcluster/frd
ump.cmcld.6 (length = 161466).
Feb 4 15:09:48 Ecards1 cmcld[3434]: Obtaining Cluster Lock
Feb 4 15:09:48 Ecards1 cmdisklockd[3445]: Obtaining cluster lock device /dev/ds
k/c4t0d1
Feb 4 15:09:48 Ecards1 cmdisklockd[3445]: Successfully obtained the lock at /de
v/dsk/c4t0d1!