cluster virtual ip address loss

[Janrum]

Hi,

We have a CL380 Proliant Cluster system operational with MS W2000 Advanced Server
Cluster services. Is has two nodes : N1 with 172.16.32.7 and N2 with 172.16.32.8.
Our cluster virtual address(VI) is 172.16.32.6. Hard beat is configured in 192 range.
We tested the redundancy on the machines and it works ok ; if n1 is down (power off)
all cluster services are taken over by n2 and VI is reachable (and vice versa).
Another test is to unplug NIC-cable N1 => N1 is not pingable, N2 is pingable but
only with its own ip address the VI is not pingable anymore ! I suppose that the services on N1 don't switch to N2 in this case and that the VI is still allocated with N1.
Is this only a configuration problem or is their more to it.

Thanks in advance for your help.

 

[tmartfeld]

This may be of help

http://support.microsoft.com/default.aspx?scid=kb;EN-US;q257925

SYMPTOMS
Cluster Server in Windows NT Server 4.0, Enterprise Edition, does not detect network problems. Disconnecting the network cable from the active node does not cause an IP address or group to failover. Also, simulated network adapter failures may not cause an IP address to stop working, or cause Cluster Server to declare that an adapter is down.

CAUSE
The Cluster service verifies that the IP address is online by checking the network stack to make sure that it is functioning. A network problem does not invalidate the IP addresses that are registered on that adapter. This causes the IP address to stay online even though the route is down.

This functionality is included in Microsoft Windows 2000 Advanced Server.

 

[fablexo]

Hi,

Try this technet article : Clustering Information on IP Address Failover (Q168567).

Maybe it could be a problem with your LAN card model or something like that.

Bye.

 

[amante]

In Win2k, if network connectivity is lost, the TCP/IP stack is unloaded, which means that all resources that depend on IP addresses are taken offline. This problem has been solved in .NET