jueves, 03 de mayo de 2012 10:12
This is kinda a theoretical question, but we are currently doing some tests on this issue, and we are having some unexpected results.
We have a 2 node windows 2008r2 cluster which is running virtual, and we are using disk and node -majority. We are NOT using a private heart-beat network. We are trying to figure out if we really need this network, and how the cluster will react without one.
Now we did the following test: The cluster group is located on clusternode test12.
We have 2 vlan's : 1 for the public network and 1 for the storage. We disabled the netwerkadapter for the public network on test12 , so as a consequence the client acces point to reach the cluster failed. However, the cluster group itself didn't failover to the other node test11.
Is this normal behavior? We thought that as a consequence the cluster group would failover, but it wasn't the case.
We also find it strange that there isn't n action in the graphical interface that allows you to move the cluster group, or adjust settings.
someone said that maybe the maximum amount of times that the cluster group can switch, was reached. I however don't know how that we can change these values for the cluster group as the graphical interface doesn't allow this.
so my questions:
Can the cluster group failover to the other node?
Is it possible to change the properties of the cluster group? (GUI or powershell?)
Todas las respuestas
jueves, 03 de mayo de 2012 13:43Moderador
Ensure that you have uncheck the “Allow cluster network communication on this network” on the network for storage, then try to disable the network adapter for public network.
By the way, if you want to move a cluster group, you can check the following blog.Windows Server 2008 - How to Move the Quorum Disk Group
- Editado Vincent HuModerator jueves, 03 de mayo de 2012 13:45
viernes, 04 de mayo de 2012 7:25
K, following scenario, because Im loosing confidence in my own reasoning
let's say I have a 2 node cluster tesclu1 and tesclu2. Node with disk majority is being used.
tesclu2 has the cluster group, and is the only node that can actually see the quorum disk(LUN).
I have 2VLAN's:
I disable clustercommunication on the VlanStor.
Now let's say that I disable one of the netwerkadpaters that locates itself with VlanPub. I tested this with disabling the netwerkadpater on tesclu1(node that doesn't own the cluster group). tesclu1 en tesclu2 won't be able to see eachother anymore.
how is quorum been reached then?
tesclu1 has only 1 vote, as it isn't owner of the cluster group and therefor can't get the vote of the witness disk
tesclu2 has 2 votes as it is owner of the cluster group and get's the extra vote of the witness disk.
tesclu1 will be removed from the cluster, as quorum is been reached by tesclu2(with witness disk).
Is this think pattern correct?
I think this is correct, but let's say that the next time i do the same thing but instead of disabling the netwerkadapter on tesclu1, i disable the netwerkadapter on tescl2(which is owner of the cluster group). How will the cluster react on this?
Will it then move over the cluster group to the other node?
So my question is, in which situations will the cluster-group failover itself?
lunes, 07 de mayo de 2012 8:00Moderador
If you disable the network adapter on tesclu1 while tesclu2 is the owner of the cluster group, the cluster group will persist on tesclu2. If you disable the network adapter on tesclu2 while tesclu2 is the owner of the cluster group, the cluster will think tesclu2 is down and move the cluster group to tesclu1.
You can also check the following guide.
Understanding Quorum Configurations in a Failover Cluster
- Marcado como respuesta Vincent HuModerator martes, 08 de mayo de 2012 8:40