[Linux-HA] I have a error when I config 3 nodes cluster.

[Alan Robertson]

huang mingyou wrote:
> hello,list.
>        I have configd a 3 nodes cluster, m2,m5 and m7. m2 run vip1,m5
> run vip2, m7 is the backup node.and the change the cib.xml.let the
> vip1 cant't run on m5 node. that all.Then,I start test.
> 1.stop the m2 node,the m7 will run vip1.
> 2.stop the m5 node ,the m7 will run vip2.
> 3.stop the m7 node ,no things happen.
> but when I stop any two nodes ,the live node will NOT run the vip1 and
> vip2.but I want it can do this.
> I check the ha-debug file.found this error .
> ---------------------
> 
> pengine[15351]: 2005/08/17_23:03:22 info:
> mask(unpack.c:unpack_config): On loss of CCM Quorum: Stop ALL
> resources
> pengine[15351]: 2005/08/17_23:03:22 info:
> mask(native.c:native_create_actions): Leave resource IPaddr_1 (m2)
> pengine[15351]: 2005/08/17_23:03:22 info:
> mask(native.c:native_create_actions): Move  resource IPaddr_2 (m5 ->
> m2)
> -----------------------
> 
> please help me,thanks!


This is exactly the correct behavior.

If you don't have a clear majority of nodes present, then you don't have 
quorum - and you don't know for sure you're not in a split-brain 
situation - with the cluster partitioned into multiple pieces by either 
scheduler bugs or by broken communication.

You _can_ override the need for quorum if you like.  But, if you do, I'd 
recommend configuring STONITH.


-- 
     Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me 
claim from you at all times your undisguised opinions." - William 
Wilberforce