hacmp 4.5 接管问题

andrewqiu

环境是AIX 5.1-02； hacmp/es 4.5；没有共享磁盘，A/B机使用远程的NFS文件系统；资源组定义ser IP和application server。
同步没有错误；应用脚本在单机环境下测试都ok，在ha环境下A/B 机start/stop hacmp都正常，应用也正常；A机takeover shutdown，B机可以接管IP并把A机的服务起起来，但是A机clstart后不能接管会自己的IP和应用，不到1分钟自己就自己下电了。我把实际的应用脚本去掉，换成空脚本，不存在这个问题。

请大家看看是怎么回事，谢谢！

+ [[ high = high ]]
+ version=1.2
+ + cl_get_path
HA_DIR=es
+ STATUS=0
+ set +u
+ [ ]
+ exit 0
HACMP Event Summary
Event: /usr/es/sbin/cluster/events/check_for_site_down hft_fund forced
Start time: Fri Jun 10 00:57:25 2005

End time: Fri Jun 10 00:57:25 2005

Action: Resource: Script Name:
----------------------------------------------------------------------------
No resources changed as a result of this event
----------------------------------------------------------------------------

Jun 10 00:57:25 EVENT START: node_down hft_fund forced

:node_down[57] [[ high = high ]]
:node_down[57] version=1.42
:node_down[58] :node_down[58] cl_get_path
HA_DIR=es
:node_down[60] export NODENAME=hft_fund
:node_down[61] export PARAM=forced
:node_down[63] UPDATESTATDFILE=/usr/es/sbin/cluster/etc/updatestatd
:node_down[72] STATUS=0
:node_down[74] [[ -z ]]
:node_down[75] EMULATE=REAL
:node_down[78] set -u
:node_down[80] (( 2 < 1 ))
:node_down[85] rm -f /tmp/.RPCLOCKDSTOPPED
:node_down[86] rm -f /usr/es/sbin/cluster/etc/updatestatd
:node_down[88] [[ forced = forced ]]
:node_down[88] [[ hft_fund = hft_fund ]]
:node_down[88] [[ REAL = EMUL ]]
:node_down[97] clchdaemons -r -d clstrmgr_scripts -t resource_locator
:node_down[102] cl_echo 3048 node_down : hft_fund forced down. Exiting immediately.\n node_down hft_fund
:cl_echo[49] version=1.10
:cl_echo[97] HACMP_OUT_FILE=/tmp/hacmp.out
Jun 10 2005 00:57:25node_down : hft_fund forced down. Exiting immediately.
:node_down[103] return 0
Jun 10 00:57:25 EVENT COMPLETED: node_down hft_fund forced

HACMP Event Summary
Event: node_down hft_fund forced
Start time: Fri Jun 10 00:57:25 2005

End time: Fri Jun 10 00:57:28 2005

Action: Resource: Script Name:
----------------------------------------------------------------------------
No resources changed as a result of this event
----------------------------------------------------------------------------

Jun 10 00:57:28 EVENT START: node_down_complete hft_fund forced

:node_down_complete[57] [[ high = high ]]
:node_down_complete[57] version=1.2.3.39
:node_down_complete[58] :node_down_complete[58] cl_get_path
HA_DIR=es
:node_down_complete[60] export NODENAME=hft_fund
:node_down_complete[61] export PARAM=forced
:node_down_complete[63] VSD_PROG=/usr/lpp/csd/bin/hacmp_vsd_down2
:node_down_complete[64] HPS_PROG=/usr/es/sbin/cluster/events/utils/cl_HPS_init
:node_down_complete[73] STATUS=0
:node_down_complete[75] [ ! -n ]
:node_down_complete[77] EMULATE=REAL
:node_down_complete[80] set -u
:node_down_complete[82] [ 2 -lt 1 ]
:node_down_complete[88] [[ forced = forced ]]
:node_down_complete[88] [[ hft_fund = hft_fund ]]
:node_down_complete[88] [[ REAL = EMUL ]]
:node_down_complete[97] clchdaemons -r -d clstrmgr_scripts -t resource_locator
:node_down_complete[102] cl_echo 3048 node_down_complete : hft_fund forced down. Exiting immediately.\n node_down_complete hft_fund
:cl_echo[49] version=1.10
:cl_echo[97] HACMP_OUT_FILE=/tmp/hacmp.out
Jun 10 2005 00:57:29node_down_complete : hft_fund forced down. Exiting immediately.
:node_down_complete[103] return 0
Jun 10 00:57:29 EVENT COMPLETED: node_down_complete hft_fund forced

HACMP Event Summary
Event: node_down_complete hft_fund forced
Start time: Fri Jun 10 00:57:28 2005

End time: Fri Jun 10 00:57:30 2005

Action: Resource: Script Name:
----------------------------------------------------------------------------
No resources changed as a result of this event
----------------------------------------------------------------------------
+ [[ high = high ]]
+ version=1.2
+ + cl_get_path
HA_DIR=es
+ STATUS=0
+ set +u
+ [ ]
+ exit 0
HACMP Event Summary
Event: /usr/es/sbin/cluster/events/check_for_site_down_complete hft_fund forced
Start time: Fri Jun 10 00:57:30 2005

End time: Fri Jun 10 00:57:30 2005

Action: Resource: Script Name:
----------------------------------------------------------------------------
No resources changed as a result of this event
-------------------------------------------------------------

matata

没碰到，一般都是在有共享盘阵的HA模式下。

wildhorse

这个日志有用信息太少，只知道是被强迫down。

mithara

你的资源模式是那种模式的呢.concrrent还是cacading?
这种类似的问题，我也遇到过，
我先说说我的情况，
1，采用rotating资源模式。双机从A切换到B时候正常，再切回来的时候，A机关机。我装的是ORACLE，根据ORACLE工程师讲，ORACLE是有调用关机的这样的命令。因为系统阻塞造成的关机。但我对这个解释持保留意见。呵呵
2、concrrent方式下。同样不切换，关机。检查一下，原来是HACMP中串口心跳没配的原因。造成双机不能通讯造成的。
不过，象第一种现象，我重启机器后，在切换，有可能就没问题了。也就是说它不是必然的，而是偶然的。
因此，我觉得
你应该多试几次。看看是必然的问题还是偶然的问题。
要仔细检查一下网络和串口通信线。
都正常的情况下，你就要看你的应用软件内部有没有什么问题，是不是有什么异常的地方了。

smartlam

可能是芯跳的原因。执行/usr/es/sbin/cluster/clstat，看看tty那行是up定down？

RS9000

1、检查你的拓扑和资源配置，确认没有问题
2、检查HACMP的补丁，确定补丁正确
3、切到B机后，先不要切回A机，而是在B机做听应用测试，有的时候是你的应用不能停造成的
总之，从配HA开始就一步步做好，要不然后来会很麻烦！