Bad ARP Broadcasts from Cluster -
03-04-2004
, 08:08 AM
Hello,
I am running a 2 node cluster with 3 instances of SQL. One of the instances,
SQLSERVER01, is used exclusively by a front end Windows Server 2003 Web
Server. Problem I am having is, when the SQLSERVER01 instance resides on a
specific node of the cluster, the web server can communicate with it just
fine. When the instance is moved to the other node in the cluster, the web
server receives the updated MAC from the clusters ARP broadcast, but after
two minutes or so it reverts back to the MAC of the original cluster node. I
can do an 'arp -d' to clear the web servers ARP cache and everything is
alright for another two minutes, when all of a sudden the worng MAC shows up
again in the web servers ARP table. I have checked other client machines on
the VLAN and they all have the correct MAC address, seems to be only
effecting the web server that is constantly communicating with the cluster.
I have checked the switch, a Cisco 3548XL, and its ARP cache shows the
correct MAC address.
Any ideas? If any clarification is needed just let me know. I was thinking
of posting this in a win2003 newsgroup but couldnt find one.
Thanks,
JBailey |