martes, 12 de junio de 2012 16:56
This is most easily seen using IIS, which I know has a process to add it to the cluster, but this is not about IIS but about its behavior. Basically, I added a bit of text to the default IIS.png to differentiate between the two servers in the cluster. I also added the WWW Publishing service to the cluster group to be started and stopped as the cluster moved from node to node. What I found was that when the default webpage was accessed using the cluster name ("Galaxy") the page displayed would be from the node that was the current quorum owner. So if one node ("Mercury") had all of the active services, but the other node ("Venus") was the quorum owner the displayed webpage would be from Venus.
The fix was simple, in an administrative command window execute: cluster group "cluster group" /move
This moved the quorum drive from Venus to Mercury and without any other changes and simply hitting refresh on the browser the displayed webpage would be from Mercury.
This behavior is important because the application that I have been tasked with clustering is behaving in a similar manner. Accessing it through the Cluster Name results in the requests being forwarded to the node with the quorum, and not the node with all of the active services.
Is there a way to change the behavior? Either by having the quorum drive move with the failover, and I haven't found a way to do that automatically, or having the cluster name forward all requests to the active node.
Windows Genuine DISAdvantage
Todas las respuestas
miércoles, 13 de junio de 2012 16:43Moderador
The best method is not to place anything on the Quorum disk and not to use the Cluster NetBIOS name in a website. these things are only for managing the cluster
What you can do is create a new resource with a NetBIOS name and a disk and use this for IIS but remember only one IIS server can access this.
Better is create a NLB for IIS with or without shared configuration that way you have 2 IIS servers serving the same website and double the uptime.
How to create a NLB cluster
Greetings, Robert Smit [MVP] http://robertsmit.wordpress.com/ “Please click "Vote As Helpful" if it is helpful for you and Proposed As Answer”
- Marcado como respuesta Cuervo_AG jueves, 14 de junio de 2012 19:46
jueves, 14 de junio de 2012 19:46
Thanks for the response. Nothing is on the quorum disk, just that using the cluster name to access the services on the active node will forward you to the node on which the quorum is owned. This behavior is extremely irritating, and in a simple active/passive 2-node cluster unnecessary. Why is it that there are 4 different selections for quorum behavior, but forwarding requests sent to the cluster name to the active node is something that isn't selectable?
We ended up using our F5 load balancer to determine which cluster node is actually owning the services. We edited the health monitor to check the ports every 10 seconds which corresponds to the amount of time it takes for the services to come up on the new node.
As it is this cluster is using 5ip addresses: the Cluster Name, the Client Access Point, Node 1, Node 2 and the NLB IP. 2003 did the same thing with 3 IPs, if I recall correctly.
Windows Genuine DISAdvantage
- Editado Cuervo_AG jueves, 14 de junio de 2012 19:51