前些天有一个朋友安装CRS的时候无法找到节点，具体错误信息如下：
$./cluvfy/runcluvfy.shstage -post hwos -n node1,node2 -verbose
notice: JAVA_HOME not setin the environment.
Performingpost-checksfor hardware and operating system setup
Checkingnodereachability...
Check:Nodereachability from node "node1"
DestinationNode Reachable?
------------------------------------ ------------------------
node1 no
node2 no
Result: Node reachabilitycheck failed from node "node1".
碰到这个问题，首先我问他hostname是否就是Node1，node2，而且这两个地址是否正确的在/etc/hosts里定义了。他回答是这些都没错，然后我问他网卡的顺序是否正确，在HP-UX下有时候私网和公网的网卡在两台服务器上的设备名不对应，也可能会出一些莫名其妙的错误。经过确认，这些配置都没问题。
接下来检查hosts文件和services文件的属性，也都没问题，world都有r的权限。看样子问题不是出在这些地方。那么下面该怎么处理呢？
在常规方法无法找到问题的时候，使用操作系统跟踪工具对某个出问题的配置过程进行跟踪是十分有效的方法，我建议他用tusc跟踪一下，看看能有什么发现。朋友是在HP-UX上安装集群软件，于是我建议他开始使用tusc去监控一下堆栈。
tusc的TRACE文件发给我以后，很快在tusc里面我们发现了疑点：
stat("/usr/sbin/ping",0x742ff5e4)....................................................................= 0
access("/usr/sbin/ping",X_OK)........................................................................ =0
pipe() ................................................................................................=9 (10)
pipe()................................................................................................=11 (12)
pipe()................................................................................................=13 (14)
vfork()............................................................ (returningas child...) ......... , 2452
execve("/tmp/2437/jre/bin/HPUXChildWrapper",0x1923d0, 0xf5230)....................................... [entry]
argv[0]@ 0x192410: "/tmp/2437/jre/bin/HPUXChildWrapper"
argv[1]@ 0x742ff460: "9"
argv[2]@ 0x742ff470: "12"
argv[3]@ 0x742ff480: "14"
argv[4]@ 0xffffffffc1be1140:
argv[5]@ 0x742ff038: "/usr/sbin/ping"
argv[6]@ 0x192170: "node1"
argv[7]@ 0x192188: "-n"
argv[8]@ 0x1921a0: "1"
argv[9]@ 0x1921e8: "-m"
argv[10]@ 0x192200: "3"
env[0]@ 0x77ff00f2: "_=/tmp/2437/jre/bin/java"
env[1]@ 0x77ff010b:"MANPATH=/usr/share/man/%L:/usr/share/man:/usr/contrib/man/%L:/usr/contrib/man:/usr/local/man/%L:/usr/local/man:/opt/mx/share/man:/opt/upgrade/share/man/%L:/opt/upgrade/share/man:/opt/pd/share/man/%L:/opt/pd/share/man:/opt/pd/share/man/%L:/opt/pd/share/man:/opt/pd/share/man/%L:/opt/pd/share/man:/opt/resmon/share/man:/opt/gnome/man:/opt/openssl/man:/opt/openssl/prngd/man:/opt/wbem/share/man:/opt/hparray/share/man/%L:/opt/hparray/share/man:/opt/graphics/common/man:/usr/dt/share/man:/opt/samba/man:/opt/perl/man:/opt/ignite/share/man/%L:/opt/ignite/share/man:/opt/ssh/share/man"
env[2]@ 0x77ff034e: "SHLIB_PATH=/tmp/2437/jre/lib/PA_RISC2.0:/tmp/2437/jre/lib/PA_RISC2.0/server:/tmp/2437/jre/../lib/PA_RISC2.0:/tmp/2437/lib32:/tmp/2437/srvm/lib32:"
env[3]@ 0x77ff03e0: "PATH=/usr/bin:/usr/ccs/bin:/usr/contrib/bin:/opt/hparray/bin:/opt/nettladm/bin:/opt/upgrade/bin:/opt/fcms/bin:/opt/pd/bin:/opt/resmon/bin:/opt/gnome/bin:/usr/bin/X11:/usr/contrib/bin/X11:/opt/mozilla:/opt/wbem/bin:/opt/wbem/sbin:/opt/graphics/common/bin:/usr/sbin/diag/contrib:/opt/mx/bin:/opt/perl/bin:/opt/ssh/bin:."
env[4]@ 0x77ff051e: "COLUMNS=125"
env[5]@ 0x77ff052a: "ORACLE_BASE=/home/oracle/base"
env[6]@ 0x77ff0548: "EDITOR=vi"
env[7]@ 0x77ff0552: "LOGNAME=oracle"
env[8]@ 0x77ff0561: "CV_DESTLOC=/tmp"
env[9]@ 0x77ff0571: "ERASE=^H"
env[10]@ 0x77ff057a: "CV_JDKHOME=/tmp/2437/jre"
env[11]@ 0x77ff0593: "CRS_HOME=/home/oracle/crsHome"
env[12]@ 0x77ff05b1: "SHELL=/sbin/sh"
env[13]@ 0x77ff05c0: "HOME=/home/oracle"
env[14]@ 0x77ff05d2: "LD_LIBRARY_PATH=/tmp/2437/lib:"
env[15]@ 0x77ff05f1: "TERM=vt100"
env[16]@ 0x77ff05fc: "CV_HOME=/tmp/2437"
env[17]@ 0x77ff060e: "PWD=/home/oracle"
env[18]@ 0x77ff061f: "TZ=EAT-8"
env[19]@ 0x77ff0628: "LINES=47"
env[20]@ 0xffffffffc1be0fa8:
env[21]@ 0xffffffffc1be0fd0:
从上述的TRACE我们看出，检查过程中调用了Ping，估计CRS安装时候使用PING命令去检查节点的健康性。从后面的参数来看，我们看到其中有-n -m等ping的参数，于是我让他执行一下/usr/sbin/pingnode1 -n 1 -m 3 。
反馈的信息让我眼前一亮，居然发现ping没有-m选项：
# ping node1 -n1
PING node1: 64 bytepackets
64 bytes from192.168.1.150: icmp_seq=0. time=0. ms
----node1PINGStatistics----
1 packets transmitted, 1packets received, 0% packet loss
round-trip (ms) min/avg/max = 0/0/0
# ping node1 -n 1 -m 3
Usage: ping [-oprv][-i address] [-t ttl] host [-n count]
ping [-oprv] [-i address] [-t ttl]host packet-size [[-n] count]
这下子就很明确了，估计是HP-UX的补丁没有打全，所以ping命令没有-m选项。经过检查，发现GOLDQPK11i 没有打，打了补丁后，测试一下/usr/sbin/pingnode1 -n 1 -m 3 ,一切都OK了，于是再安装CRS，一切正常。
这个案例主要让大家学会如何通过TUSC去做深入的诊断，如果发现问题后胡乱的从网上去找解决方案，肯定是不行的。解决问题，思路最重要。
