找了好久都没找到的HA解决方法，来这里碰碰运气看

atianyu

环境：两台RH3，做HA，软件是自带的
现在情况是，集群已经做好，添加进去的系统自带服务器HTTPD也能运行，但是想从rh3-ha1切换到rh3-ha2时，总是不成功，状态显示disable
说这么多，先看看图吧
rh3-ha2上的状态查询
Cluster Status - rh3-cluster                                           16:23:00
Cluster Quorum Incarnation #1
Shared State: Shared Raw Device Driver v1.2

  Member             Status   
  ------------------ ----------
  192.168.1.120      Active               
  192.168.1.121      Active     <-- You are here

  Service        Status   Owner (Last)     Last Transition Chk Restarts
  -------------- -------- ---------------- --------------- --- --------
  httpd          disabled (None)           04:13:30 Jan 04   4        0

rh3-ha1上的状态查询
Cluster Status - rh3-cluster                                           05:31:22
Cluster Quorum Incarnation #1
Shared State: Shared Raw Device Driver v1.2

  Member             Status   
  ------------------ ----------
  192.168.1.120      Active     <-- You are here
  192.168.1.121      Active               

  Service        Status   Owner (Last)     Last Transition Chk Restarts
  -------------- -------- ---------------- --------------- --- --------
  httpd          started  192.168.1.120    05:18:07 Jan 04   4        0

然后我是想把rh3-ha1的httpd停掉后，rh3-ha2的状态会变成started,而rh3-ha1的状态会变成stoped
结果却不是，如下
[root@rh3-ha1 log]# clusvcadm -s httpd -m 192.168.1.120
Member 192.168.1.120 stopping httpd...success

Cluster Status - rh3-cluster                                           05:33:46
Cluster Quorum Incarnation #1
Shared State: Shared Raw Device Driver v1.2

  Member             Status   
  ------------------ ----------
  192.168.1.120      Active     <-- You are here
  192.168.1.121      Active               

  Service        Status   Owner (Last)     Last Transition Chk Restarts
  -------------- -------- ---------------- --------------- --- --------
  httpd          stopped  (192.168.1.120)  05:33:18 Jan 04   4        0



rh3-ha2上
Cluster Status - rh3-cluster                                           16:26:22
Cluster Quorum Incarnation #1
Shared State: Shared Raw Device Driver v1.2

  Member             Status   
  ------------------ ----------
  192.168.1.120      Active               
  192.168.1.121      Active     <-- You are here

  Service        Status   Owner (Last)     Last Transition Chk Restarts
  -------------- -------- ---------------- --------------- --- --------
  httpd          disabled (None)           04:13:30 Jan 04   4        0

然后想再手动去在rh3-ha2 来启动这个服务，结果如下
[root@rh3-ha2 init.d]# clusvcadm -e httpd -m 192.168.1.121
Unknown service: httpd
[root@rh3-ha2 init.d]#

下面传两张这两台设备的图形界面，其中30那张是rh3-ha2的，31那张是rh3-ha1的（因为我手动停止了httpd服务，所以这里显示的是停止）
现在问题就是为什么rh3-ha2就无法接管服务？

[ 本帖最后由 atianyu 于 2010-1-4 16:32 编辑 ]

atianyu

为什么没人回答呢？太容易了，不屑于回答吗?自己暖和暖和吧

对于这个问题有些进展，经过自己的摸索，目前两者之间是可以切换的，也就是说，我把rh3-ha1停掉，服务是可以转移到rh3-ha2的，但是超级慢，我用PING来测试，发现当停掉ha1时在超时很久后，在能通，也就是说转移到ha2了，192.168.1.122是两台虚拟出来的对外提供服务IP。
不知道是不是哪里设置错误了，我那个服务器探测等待时间是默认的15秒。
请高手们支个招吧。
Reply from 192.168.1.122: bytes=32 time<1ms TTL=64
Reply from 192.168.1.122: bytes=32 time<1ms TTL=64
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Reply from 192.168.1.122: bytes=32 time<1ms TTL=64
Reply from 192.168.1.122: bytes=32 time<1ms TTL=64

[ 本帖最后由 atianyu 于 2010-1-5 09:12 编辑 ]

q1208c

调整一下 各个 time 的值试试.

好久不搞 rhcs 了. 不记得有哪些参数了.

不过, 注意不要调得太小. 否则会产生震荡的.

atianyu

以下是我的配置参数
[root@rh3-ha1 etc]# more cluster.xml
<?xml version="1.0"?>
<cluconfig version="3.0">
  <clumembd broadcast="yes" interval="500000" loglevel="5" multicast="no" multicast_ipaddress="" thread="yes" tko_count="20"/>
  <cluquorumd loglevel="5" pinginterval="1" tiebreaker_ip=""/>
  <clurmtabd loglevel="5" pollinterval="4"/>
  <clusvcmgrd loglevel="5"/>
  <clulockd loglevel="5"/>
  <cluster config_viewnumber="21" key="f22ec73099984401e3330bb7d713d49a" name="rh3-cluster"/>
  <sharedstate driver="libsharedraw.so" rawprimary="/dev/raw/raw1" rawshadow="/dev/raw/raw2" type="raw"/>
  <members>
    <member id="0" name="192.168.1.120" watchdog="yes">
    </member>
    <member id="1" name="192.168.1.121" watchdog="yes"/>
  </members>
  <services>
    <service checkinterval="4" failoverdomain="fd" id="0" name="httpd" userscript="/etc/rc.d/init.d/httpd">
      <service_ipaddresses>
        <service_ipaddress broadcast="192.168.1.255" id="0" ipaddress="192.168.1.122" netmask="255.255.255.0"/>
      </service_ipaddresses>
      <device id="0" name="/dev/sdc1" sharename="">
        <mount forceunmount="yes" fstype="ext3" mountpoint="/app/" options="rw"/>
      </device>
    </service>
  </services>
  <failoverdomains>
    <failoverdomain id="0" name="fd" ordered="no" restricted="no">
      <failoverdomainnode id="0" name="192.168.1.121"/>
      <failoverdomainnode id="1" name="192.168.1.120"/>
    </failoverdomain>
  </failoverdomains>
</cluconfig>

q1208c

呵呵. 我都看不懂了.
你看一下手册. 那个图形的管理器上好象就可以配置的.

zhang1980s

楼主，虽然看上去你那个问题应该有解，但是RHEL3今年就要退出支持了，因此实在没兴趣再架环境搞RHEL3的HA，你还是考虑换个高版本RHEL的HA把。红帽早就抛弃RHEL3HA的架构了，高版本的RHCS会好用很多。

jerrywjl

既然能failover，证明这个功能应该是OK的。但为什么这么慢，还是要跟踪日志。
如果日志也没有什么问题，那么interval和tko是不是太长了点？

但反正正如LS说，RHEL3上的RHCS已经过气了。对此我也不是很熟。

atianyu

这位兄弟说得很中肯呀，非常赞同，以前我做Novell公司的SUSE HA的时候，那功能就强大，而且非常人性化的配置，用起来爽呀。
我主要是想多接触点看看，其实当初是想用redhat 5的HA玩玩，连系统都装好了，环境也准备了，就是没那些集群方面的软件，不知道兄弟你有吗，共享 下看看。

atianyu

谢谢回答，回去试试改些和时间有关的东西，NND，就不信测不出来

atianyu

诶呀，各位高手，还是不行，改了一些时间缩短，还是一样很长时间才能切换过来
很奇怪的就是这rh3的cluster里怎么就没有设置心跳IP地址的地方？
按照安装指导的说法，要在两台机器的/etc/hosts写类似10.0.0.1     xintiao1
                                                                            10.0.0.2     xintiao2

但在实际的配置的时候貌似根本没用到这两个IP，是不是多余的呀？