HACMP5.2不切换问题，急！！！

longzhiyue

机型为：P550
双机配置为冷备份，先在SDP1上拉起cluster，双机的三个进程都起来了，然后共享磁盘也激活了，数据库和应用都正常起来，然后再去SDP2起HA，它的三个进程也都启动正常。当我用smit clstop/takeover在停本节点SDP1，请求SDP2进行接管，可是SDP1上的数据库和应用都没有停，HA的三个进程中停了两个，还有clstrmgrES一直是stopping状态。麻烦各位前辈们看看！

附件是hacmp.out的日志

下面贴上日志：event.log
===============================================
Fri May 25 21:29:28 TAIST 2007
Fataling Device : Network
Scription : event : network_down
nodename : -1
network_name: rsnet

Remote node or global network down.

[ 本帖最后由 longzhiyue 于 2007-5-26 23:33 编辑 ]

chinadns

双机监控脚本配置的有点问题，就是scppkg包对应的双机脚本，仔细看看

kill 应用能不能切换，正常的话应该是可以切换的

longzhiyue

不好意思日志贴得有问题，现在整个就是这样一个情况，当我KILL应用的时候，应用会自动被监控拉起来，双机不会切换。当用smit clstop/takeover时，应用和数据库一直还在，clstrmgrES一直stopping。

这是spy.logMay 26 12:18:42 - Node "sdp1": Start all customer's commands beginning...
May 26 12:18:42 - Node "sdp1": Start informix ...
May 26 12:18:43 - Node "sdp1": Informix start now.
May 26 12:19:01 - Node "sdp1": CDR_SUPPORT is set to N, so do not start cdr server .
May 26 12:19:01 - Node "sdp1": Start informix completed.
May 26 12:19:01 - Node "sdp1": oninit[informix] start successful.
May 26 12:19:01 - Node "sdp1": Start sdu application ...
May 26 12:19:01 - Node "sdp1": Sdu now starting ...
May 26 12:19:07 - Node "sdp1": Start sdu application completed.
May 26 12:19:07 - Node "sdp1": manager[sdu] start successful.
May 26 12:19:07 - Node "sdp1": Start all customer's commands complete.
May 26 12:19:11 - Node "sdp1": Dual system mode = primary.
May 26 12:19:11 - Node "sdp1": Monitor loop[0] begin.
May 26 12:19:11 - Node "sdp1": Monitor - oninit[informix] total pid num=1, list=21064
May 26 12:19:11 - Node "sdp1": Monitor - manager[sdu] total pid num=1, list=8292
May 26 12:19:11 - Node "sdp1": Total application number: 2
May 26 12:19:11 - Node "sdp1": Monitored applications/processes by system: 2
May 26 12:19:11 - Node "sdp1": Monitored applications/processes by user: 0

event.log:===============================================
Sat May 26 12:18:47 TAIST 2007
Fataling Device : Network
Scription : event : network_down
nodename : -1
network_name: rsnet

Remote node or global network down.

longzhiyue

弱弱的问下，双机调试能用两根交叉网线把两台小型机连接起来吗？

chinadns

两根交叉网线连起来干什么？

还是把双机脚本好好看一遍，看看是不是哪里配错了，在ha里还需要配置一个NOTIFY ，这个很重要，直接影响切换，

longzhiyue

不切换的原因是双机通信不正常，在主机用clstat -a看到备机的状态是DOWN的，当在备机用此命令的时候全是DOWN，但是我的主备机HA是起来了的啊，三个进程都是active状态，我感觉最主要还是在解决这个双机的通信问题，原来是snmpdv3ne，我把它改成snmpdv1试过，通信还是不正常。

因HA5.2是用rhosts通信，/usr/es/sbin/cluster/etc/rhosts这个文件是这样配置的不知道正确不？
sdp1
sdp1_svc
sdp1_stb
sdp2
sdp2_svc
sdp2_stb
两台机子上我都是这么配置的

下面是hosts配置的内容，两台一样。
# sdp1
172.31.1.21               sdp1_svc        sdp1
172.32.1.21               sdp1_stb

# sdp2
172.31.1.22               sdp2_svc        sdp2
172.32.1.22               sdp2_stb

# services ip address(share ip address)
172.30.1.23             share_ip

[ 本帖最后由 longzhiyue 于 2007-5-27 19:22 编辑 ]

jtw

这个不需要手动设置

jtw

你的rs232网络有没有问题

longzhiyue

rs232网络应该没有问题

我在sdp1#cat < /dev/tty0
然后sdp2#cat /etc/hosts > /dev/tty0后，SDP1上可以正确显示/etc/hosts的内容，在SDP2上也同样这样试过

双机是启动状态下
我在执行下面操作的时候显示的信息
HACMP Extended Resources Configuration
Configure HACMP Applications
Show Cluster Applications

当双机在两个节点没有启动的时候再执行就得不到这样的输出了，只有no respond

我使用odmget HACMPadapter是可以显示我配置的双机网络信息。
======================Show Cluster Applications输出===================
WARNING: Unable to locate interface 'en0' on node 'sdp1'.

WARNING: An unknown problem occurred while executing the ip_of_interface subroutine.

ERROR: Unable to find the network 'ethernet' in the Management Information Base.

WARNING: Unable to locate interface 'en2' on node 'sdp1'.

WARNING: An unknown problem occurred while executing the ip_of_interface subroutine.

ERROR: Unable to find the network 'ethernet' in the Management Information Base.

WARNING: Unable to locate interface 'en0' on node 'sdp2'.

WARNING: An unknown problem occurred while executing the ip_of_interface subroutine.

ERROR: Unable to find the network 'ethernet' in the Management Information Base.

WARNING: Unable to locate interface 'en2' on node 'sdp2'.

WARNING: An unknown problem occurred while executing the ip_of_interface subroutine.

ERROR: Unable to find the network 'ethernet' in the Management Information Base.

WARNING: Could not determine the state of application 'scppkg'.

WARNING: Could not determine the state of application scppkg.
Cluster: sdpcluster
   Cluster services: active

WARNING: translate() was called on the word '' which is not in translation table.
   State of cluster: 1

WARNING: translate() was called on the word '' which is not in translation table.
      Substate: 1

#############
APPLICATIONS
#############
   Cluster sdpcluster provides the following applications: scppkg  

WARNING: translate() was called on the word '' which is not in translation table.
      Application: scppkg {1}
         This application is part of resource group 'sdp_res'.
            Resource group policies:
               Startup: on first available node
               Fallover: to next priority node in the list
               Fallback: never
         Nodes configured to provide scppkg:  
WARNING: translate() was called on the word '' which is not in translation table.
sdp1 {1}   
WARNING: translate() was called on the word '' which is not in translation table.
sdp2 {1}   
         scppkg is started by /usr/sbin/cluster/utilities/tellin/scp/scppkg/start_scp.sh
         scppkg is stopped by /usr/sbin/cluster/utilities/tellin/scp/scppkg/stop_scp.sh
         Resources associated with scppkg:
            Service Labels
               share_ip(172.30.1.23)
                  Interfaces configured to provide share_ip:
                     sdp1_svc
                        with IP address: 172.30.1.21
                        on interface: en0
                        on node: sdp1
                        on network: ethernet
                     sdp1_stb
                        with IP address: 172.31.1.21
                        on interface: en2
                        on node: sdp1
                        on network: ethernet
                     sdp2_svc
                        with IP address: 172.30.1.22
                        on interface: en0
                        on node: sdp2
                        on network: ethernet
                     sdp2_stb
                        with IP address: 172.31.1.22
                        on interface: en2
                        on node: sdp2
                        on network: ethernet
            Shared Volume Groups:
               sdpvg
               datavg
            Filesystems:
               /tellin/datasync
               /tellin/sdu
               /tellinshare/sms
         No application monitors are configured for scppkg.

#############
TOPOLOGY
#############
   sdpcluster consists of the following nodes: sdp1 sdp2  
      sdp1
         Network interfaces:
            sdp1_svc
               with IP address: 172.30.1.21
               on interface: en0
               on network: ethernet
            sdp1_stb
               with IP address: 172.31.1.21
               on interface: en2
               on network: ethernet
      sdp2
         Network interfaces:
            sdp2_svc
               with IP address: 172.30.1.22
               on interface: en0
               on network: ethernet
            sdp2_stb
               with IP address: 172.31.1.22
               on interface: en2
               on network: ethernet

chinadns

这个系统不说装烂了，至少也装了十几二十遍了，都已经实现了脚本自动化安装，
从系统到informix安装到双机配置，脚本跑下来，2个小时不到

ERROR: Unable to find the network 'ethernet' in the Management Information Base.
网络名应该是ethernet01吧