G
Guest
Here's an obscure one that any help would be greatly appreciated with:
We just recently built a 2000 Advanced failover cluster and the resources
that are provided by the cluster service failover perfectly when we bring one
of the two nodes down. The issue that we're having is, some of our remote
branches can see the resources on the cluster and some can't. The common
denominator is, the branches that can see the cluster resources are separated
by routers, and the ones that can't, are separated by Cisco Catalyst 2948G-L3
switches, which primarily does layer 2 switching but can do layer 3.
What makes it even more odd is, when you ping the virtual resource on the
cluster from the local Cisco switch (as in, logged into the Cisco IOS which
is 12.0), all ping packets fail! No matter where you are in our system
though (different subnets separated by switches as well), you can always ping
the two nodes that host the resources. The only way we have been able to get
the remote branches that have switches between them to see the cluster
resources, is by adding a static arp entry into the local switch with the MAC
address of the current owner of the resource. Since the switch won't allow
two different MAC entries pointing to one IP address, for obvious reasons, if
the cluster fails over connectivity is lost defeating the purpose of the
cluster.
I've found articles from MS KB talking about this type of issue but it
involves load balanced clusters, but have still tried some of their
suggestions without any luck.
My forehead is getting lumpy and the wall is becoming dented and thus the
reason I've turned to the experts.
Thanks in advance everyone,
Rick
We just recently built a 2000 Advanced failover cluster and the resources
that are provided by the cluster service failover perfectly when we bring one
of the two nodes down. The issue that we're having is, some of our remote
branches can see the resources on the cluster and some can't. The common
denominator is, the branches that can see the cluster resources are separated
by routers, and the ones that can't, are separated by Cisco Catalyst 2948G-L3
switches, which primarily does layer 2 switching but can do layer 3.
What makes it even more odd is, when you ping the virtual resource on the
cluster from the local Cisco switch (as in, logged into the Cisco IOS which
is 12.0), all ping packets fail! No matter where you are in our system
though (different subnets separated by switches as well), you can always ping
the two nodes that host the resources. The only way we have been able to get
the remote branches that have switches between them to see the cluster
resources, is by adding a static arp entry into the local switch with the MAC
address of the current owner of the resource. Since the switch won't allow
two different MAC entries pointing to one IP address, for obvious reasons, if
the cluster fails over connectivity is lost defeating the purpose of the
cluster.
I've found articles from MS KB talking about this type of issue but it
involves load balanced clusters, but have still tried some of their
suggestions without any luck.
My forehead is getting lumpy and the wall is becoming dented and thus the
reason I've turned to the experts.
Thanks in advance everyone,
Rick