Re: Network cluster problems

From: John Toner [MVP] (jtoner_at_mvps.DIE.SPAM.DIE.org)
Date: 05/11/04 

Date: Tue, 11 May 2004 13:26:30 -0400

First, it is recommended to disable media sense on the "heartbeat" network
connection, not necessarily the public connection. If you disable it on the
public, W2K will no long failover if you lose network connectivity. Some
people (like myself) prefer to have media sense disabled on the public
connections as I would rather have "teaming" control network failover rather
than MSCS, as MSCS doesn't do a good job restarting these resources once
they've failed...typically causing your applications to remain offline. So,
if you want MSCS to failover during a network failure, you must leave the
DisableDHCPMediaSense value set to zero.

For your second issue, when you lose network connectivity the IP resources
in the cluster will fail taking down any network names and other resource
dependant upon these resources. You're trying to connect to the cluster
using the "Network Name" value of the cluster and this resource should be in
an offline state...hence the reason you're getting the 1722 error. To
re-connect to Cluster Administrator, simply enter a period (.) instead of
the cluster name to connect to the cluster using a local procedure call
instead of RPC. This will force Cluadmin to connect to the local cluster
service (if available) and you'll then be able to bring the IP resource back
online.

Here's some additional articles that might help to explain this behaviour:

http://support.microsoft.com/default.aspx?scid=kb;en-us;814459
http://support.microsoft.com/default.aspx?scid=kb;EN-US;242600

Hope this helps.

Regards,
John

"Jose" <anonymous@discussions.microsoft.com> wrote in message
news:b43a01c43749$4487b060$a501280a@phx.gbl...
> I wrote some time ago about a problem I usually have with
> Microsoft Cluster, but I had problems to make me
> understood (I know how my English is). I am installing a
> new cluster and I going to try to explain all the problems
> I have detected. My cluster is formed by:
> Node 1: 2 hard disks forming RAID 1. Operating system,
> Node 2: 2 hard disks forming RAID 1. Operating system,
> Shared disks: 2 hard disks forming RAID 1 as quorum
> 3 hard disks forming RAID 5 for database.
> Each node has two network adapters-one for connection to
> the public network and the other for the node-to-node
> private cluster network.
> The first test I made it was to take out the public
> network cable of the node where the cluster was up. But
> the cluster didn't failover. I read an article about the
> DHCPMedia Sense parameter
> (HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Tcpip
> \Parameters\DHCPMediaSense) in the registry that it had to
> be 1. I checked the registry and it was 1, but I changed
> it to 0 and now when I take out the public network cable
> it failovers. All the resources change to the other node
> and the public network connection disappears from the
> server in the Microsoft Cluster Manager. Is it right?
> Should this parameter be 0? Should the public network
> interface disappear or should it be shown as down?
>
> Other tests I have made are to take out both public
> network cables from both nodes. Obviously the cluster goes
> down. But when I put both cables again, it doesn't go up.
> If I try to open the Cluster Manager from one node I get
> this error :
> "Error trying to open the cluster in 'CLUSTER_NAME'. RPC
> server is not available. Error id: 1722 (000006ba)"
> I have to restart the node that had the cluster control,
> then it takes the control of the cluster, and then the
> cluster is up and I can manage the cluster form both
> nodes. If I restart the node that didn't have the cluster
> control I don't get the cluster up
>
>
> Thanks a lot for reading all of this with my English :-)
>

Relevant Pages

  • Re: Intermitant XP Network Problems
    ... they just can't connect resources to each other. ... the list of network PC's, and both PC's are always visible. ... I have a Win2K domain server that 2 XP Pro PC's log into. ... it will lose it's connection again some time in the future ...
    (microsoft.public.windowsxp.network_web)
  • Re: Network cluster problems
    ... >connection, ... W2K will no long failover if you lose network ... the IP resources ... >in the cluster will fail taking down any network names ...
    (microsoft.public.windows.server.clustering)
  • problem adding second node to win2k3 cluster
    ... I created a windows 2003 cluster on an ESX 2.5 box with only one node ... But I am not able to access the cluster administrator (open connection ... valid TCP/IP configurations and are connected to the network. ...
    (microsoft.public.windows.server.clustering)
  • Re: IP Address resource does not come up after failover!
    ... That's not the way cluster works. ... addresses - it brings them online provided the network supporting the IP ... switches, they are on the same VLAN? ... connection between the NICs of both machines and the switches (links are ...
    (microsoft.public.windows.server.clustering)
  • Re: Cluster does not restart SQL group after network failure
    ... Unfortunately, until the network resource is restored, no node ... including the Network Name and the SQL Server. ... cluster will bring the system online. ... > virtual SQL and Windows IP address resources went down and did not come up ...
    (microsoft.public.sqlserver.clustering)
Loading