论坛徽章:: 0

电梯直达

1楼 [收藏(0)] [报告]

发表于 2008-09-09 19:58 |只看该作者 |倒序浏览

一台机器优于温度过高, 宕机了, HACMP 产生 forced events, 没有切换 \r\n\r\n想问下, HACMP 什么情况下会产生force 事件而不产生 takeover 事件??\r\n\r\n怎么设置才能当一台宕机后, HACMP自动产生takeover 自动切换到备机上?? 手动 takeover 切换没问题的!!\r\n\r\n\r\n谢谢!!!\r\n\r\nHACMP Event Summary\r\nEvent: /usr/es/sbin/cluster/events/check_for_site_down p550a forced \r\nStart time: Sun Sep 7 09:43:52 2008\r\n\r\nEnd time: Sun Sep 7 09:43:52 2008\r\n\r\nAction: Resource: Script Name:\r\n----------------------------------------------------------------------------\r\nNo resources changed as a result of this event\r\n----------------------------------------------------------------------------\r\n\r\nSep 7 09:43:52 EVENT START: node_down p550a forced\r\n\r\n:node_down[79] [[ high = high ]]\r\n:node_down[79] version=1.45.1.1\r\n:node_down[80] :node_down[80] cl_get_path\r\nHA_DIR=es\r\n:node_down[82] export NODENAME=p550a\r\n:node_down[83] export PARAM=forced\r\n:node_down[85] UPDATESTATDFILE=/usr/es/sbin/cluster/etc/updatestatd\r\n:node_down[94] STATUS=0\r\n:node_down[96] [[ -z ]]\r\n:node_down[97] EMULATE=REAL\r\n:node_down[100] set -u\r\n:node_down[102] (( 2 < 1 ))\r\n:node_down[107] rm -f /tmp/.RPCLOCKDSTOPPED\r\n:node_down[108] rm -f /usr/es/sbin/cluster/etc/updatestatd\r\n:node_down[110] [[ forced = forced ]]\r\n:node_down[110] [[ p550a = p550a ]]\r\n:node_down[110] [[ REAL = EMUL ]]\r\n:node_down[119] clchdaemons -r -d clstrmgr_scripts -t resource_locator\r\n:node_down[124] cl_echo 3048 node_down : p550a forced down. Exiting immediately.\\n node_down p550a\r\n:cl_echo[49] version=1.13\r\n:cl_echo[98] HACMP_OUT_FILE=/tmp/hacmp.out\r\nSep 7 2008 09:43:52 node_down : p550a forced down. Exiting immediately.\r\n:node_down[125] return 0\r\nSep 7 09:43:53 EVENT COMPLETED: node_down p550a forced 0

yddll

荣誉版主

论坛徽章:: 1

2楼 [报告]

发表于 2008-09-09 22:31 |只看该作者

hacmp在机器关机的时候执行不执行接管动作，在/usr/es/sbin/cluster/etc/rc.shutdown里是会有定义的，可能你的环境问题导致的宕机应验其中了的一种情况。\r\n\r\n附/usr/es/sbin/cluster/etc/rc.shutdown部分内容：\r\n

###############################################################################\r\n# We\'re doing a shutdown because of an EPOW condition\r\n###############################################################################\r\nif [[ $EPOW_SHUTDOWN == true ]] ; then\r\n ###########################################################################\r\n # We attempt to do a \'graceful with takeover\' shutdown, so that the\r\n # resource groups get taken over when this node fails. Note that this\r\n # is in no sense guaranteed to complete before the node dies, but at\r\n # least the other cluster nodes should perform takeover.\r\n ###########################################################################\r\n clstop -gr -y -s -N -S\r\n###############################################################################\r\n# A \"fast shutdown\" (\"-F\") was specified. We turn this into a forced down\r\n# operation, mostly to preserve compatabilty\r\n###############################################################################\r\nelif [[ $fast == true ]] ; then\r\n clstop -f -y -s -N -S\r\n###############################################################################\r\n# A \"reboot\" (\"-r\") was specified. Do a graceful down, without takeover,\r\n# leaving the application cleanly down and HACMP out of the way on reboot.\r\n###############################################################################\r\nelif [[ $reboot == true ]] ; then\r\n clstop -g -y -s -N -S\r\n###############################################################################\r\n# Every other shutdown is mapped to a graceful with takeover\r\n###############################################################################\r\nelse\r\n clstop -gr -y -s -N -S\r\nfi\r\n###############################################################################\r\n# this is shutdown so remove the cluster shared memory\r\n###############################################################################\r\nsleep 4\r\nrm -f /usr/es/sbin/cluster/etc/ClSm 2>/dev/null

复制代码