redhat 5.4 cluster fence时不断发出fence指令，被fence机器不断重启

晶都小帅

dell R710
redhat 5.4 cluster ，双机热备
fence_ipmilan作为fence device

当a节点脱机时，b节点发出fence指令，a节点成功被关，a正启动时，突然又黑屏重启。查看/var/log/messages发现如下：
Jul 21 20:13:23 node3 fenced[7672]: fencing node "node4.cn"
Jul 21 20:13:53 node3 fenced[7672]: agent "fence_ipmilan" reports: Rebooting machine @ IPMI:10.0.0.44...ipmilan: Power still off Failed
Jul 21 20:13:53 node3 fenced[7672]: fence "node4.cn" failed
Jul 21 20:13:58 node3 fenced[7672]: fencing node "node4.cn"
Jul 21 20:14:03 node3 fenced[7672]: agent "fence_ipmilan" reports: Rebooting machine @ IPMI:10.0.0.44...Failed
Jul 21 20:14:03 node3 fenced[7672]: fence "node4.cn" failed
Jul 21 20:14:08 node3 fenced[7672]: fencing node "node4.cn"
Jul 21 20:14:27 node3 fenced[7672]: agent "fence_ipmilan" reports: Rebooting machine @ IPMI:10.0.0.44...ipmilan: Power still off Failed
Jul 21 20:14:27 node3 fenced[7672]: fence "node4.cn" failed
Jul 21 20:14:32 node3 fenced[7672]: fencing node "node4.cn"
Jul 21 20:15:01 node3 fenced[7672]: agent "fence_ipmilan" reports: Rebooting machine @ IPMI:10.0.0.44...ipmilan: Power still off Failed
Jul 21 20:15:01 node3 fenced[7672]: fence "node4.cn" failed
Jul 21 20:15:06 node3 fenced[7672]: fencing node "node4.cn"
Jul 21 20:15:35 node3 fenced[7672]: agent "fence_ipmilan" reports: Rebooting machine @ IPMI:10.0.0.44...ipmilan: Power still off Failed
Jul 21 20:15:35 node3 fenced[7672]: fence "node4.cn" failed


想问一下，是不是某个参数值需要配置，等待power on的时间太短所致呢？还是有其他原因？
求大牛们帮忙帮忙。。

jerrywjl

连个配置文件都没有嘛？

HalfCamel

把cluster.conf贴出来

newjava

回复 1# 晶都小帅

fence动作返回fail必然隔几十秒再重试一次；
实际fence功能实现，但是返回fail通常是fence工具的问题，可能需要通过更新fence_tool解决，不过只是尝试而已。
实际环境中部署过不少R710+RHCS5.5的集群，建议你不要用太低的版本，当前5.8都有了。