- Switches / Hubs
- DHCP
- Hyper-V
- Networking
- Network Architecture
Why does swapping a Static IP not work?
I have a Windowns NLB Cluster with a static IP. When I shutdown the NLB Cluster and give the IP to a Virtual IP on a Kemp LoadBalancer it doesn't work. I can use the same configuration for a new IP and everything works perfectly in the Kemp LoadBalancer. When I run a tracert on old IP, it goes to Kemp LoadBalancer but takes a long time -- series of Time Outs. I'm not a networking person -- why would a changed IP not work, when a new IP will? Thanks in advance.
Windows Clustering uses a weird phenomenon.
It NEVER responds to an arp request with a real MAC address. for the virtual IP... this causes the switch to broadcast it to ALL ports in the VLAN.
So if you configure one system with the real IP all traffic will always goto to THAT port only. ==> no access.
Windows Clustering cannot work without that method. (The cluster members need to coordinate between the nodes which one will answer which requests).
It NEVER responds to an arp request with a real MAC address. for the virtual IP... this causes the switch to broadcast it to ALL ports in the VLAN.
So if you configure one system with the real IP all traffic will always goto to THAT port only. ==> no access.
Windows Clustering cannot work without that method. (The cluster members need to coordinate between the nodes which one will answer which requests).
Thanks for your reply. Sorry, but I don't understand. the Windows NLB is off the network, and then steal the IP and add to a Kemp LoadBalancer as the Virtual IP. It only communicates very slows to the static IP then. If I disconnect the Nics of the old servers(Vmware) (with Microsoft NLB and static IP) - it wouldn't know it existed? Can you elaborate - thanks again. Does something/somehow still know about the old cluster?
The cluster is physically disconnected or completely off-line/shutdown?... In that case its address can be safely moved to another device.
The device needs to be in the same (V)LAN obviously.. with correct setup of tagged/untagged ports .
Microsoft clusters being a special setup might also be supported by specific configuration on switches and routers so the NLB cluster once ran in a separate small (V)LAN if only to limit the broadcast that would happen in such a LAN.
So you will need to check all systems along the path if there is special handling for that address.
The device needs to be in the same (V)LAN obviously.. with correct setup of tagged/untagged ports .
Microsoft clusters being a special setup might also be supported by specific configuration on switches and routers so the NLB cluster once ran in a separate small (V)LAN if only to limit the broadcast that would happen in such a LAN.
So you will need to check all systems along the path if there is special handling for that address.
I think that is a possibility. Thanks for thinking of it - maybe special switch setting - I'll check now.
Explore More Content
-
![]()
SolutionIs it better to purchase a larger Public IP block or consolidate current IP's in order for new modem to work with IBGP failover? -
![]()
ArticleDesign, configure and operate the IT infrastructure for a SMB (or a startup group) - Part 1 -
![]()
VideoNetwork monitoring: why having a policy is the best policy? -
![]()
Cloud Class®Certification: CCNP Cisco Certified Network Professional - Implementing Cisco IP Routing
Explore More ContentExplore courses, solutions, and other research materials related to this topic.
-
![]()
received a Solution
Can not obtain IP address -
![]()
wrote an Article
![]()
Deleting the VTP Configuration From a Cisco Switch
-
![]()
created a Video
![]()
Network monitoring: why having a policy is the best policy?
-
Explore Cloud Class® ![]()
Certification: CCNP Cisco Certified Network Professional - Troubleshooting and Maintaining Cisco IP Networks
Premium Content
You need an Expert Office subscription to comment.Start Free Trial