Troubles Clustering Windows 2003

This article is from Microsoft, it is in-the technicians that work with clusters in real world situations. In Phoenix Synergy we regularly contacted to help local businesses to (small and large) with a cluster help they need. In this case, the customer had 7 windows server 2003 from Dell, each with the standard dual Broadcom network card with 1-U server comes in those days. As you know, may be the network interfaces "Teamed" from a single interface. The customer wantstheir existing production environment as a fault tolerant as possible. The 7 new servers are new and setup in a test environment. Each Windows 2003 Standard.

What we have to work:

They have two dedicated domain controller for Active Directory, they run the DNS for the internal and external name resolution. Your domain zone records will be hosted here, they will be ns1 and ns2.

You have two web servers, and threecomplus servers. Later, they will implement their SQL cluster, but we used to people in here.

So far, it is a straightforward configuration. You want to cooperate in the network after NIC1 connected from each server switch-1 and NIC2 stuck switch-2. Taking account of a change. They will allow a cross-over cable between the two switches either NIC doomed to failure. Each switch can be connected to its own firewall / router, but the gateway on each serverSet primarily on the FW, that their switch-1 is in place, we add a second IP gateway with a different metric for each failure of the primary firewall to allow. Any firewall is connected to another provider and has a different external IP to be configured. This allows ns2 than one IP to ISP-2, which allows for a complete breakdown of the first ISP. Through all the host entries on ns2 regard to IP's from the second ISP permits the complete failure of an entire segment of theirLine.

This is the layout. Once we receive, AD and DNS setup team and we will configure the setup of the NIC. Drag a few plugs to the theory of the configuration test, and we are confident that everything goes well. So now we have to set up and test the cluster.

The cluster:

Since we do not have a network load balancer, we need to load on Web servers and server by complus Microsoft Network Load Balancing balance. We go with the NTLB management interface for clusters --the Web server. Each of the two servers in the cluster converge without incident. If we try the same on the complus servers, it does not go so well.

The problem:

We add complus1 to the cluster. It adds fine, of course, doing it the only member of the cluster. It's called "convergence" for a moment and then goes green. We try to add complus2 the cluster, and she says: "Convergence" forever, it never converges. It remains in a state of "convergence" for over 30Minutes to update after renewing stopping and starting, stopping, and tried everything. We can not converge to the second node. We try complus3 and the same result. We trace our steps, checking DNS for the internal resolution of both the server itself and the cluster IP's, all of which still looks good. We try to ping all nodes, everyone seems fine with any other ping. IPConfig shows the complus cluster IP on each server. NTLB is on each "team" interface connected. SearchMicrosoft support it that it is a problem with the NIC. So we go and try unteam each network separately. As we retrace our steps, we find the same problem regardless of how the NIC, we rely on each system. On a whim, we cancel each restart, and then add complus2 first. Then we add complus3 to the cluster. And converge "them within seconds. The attempt to add complus1 fails. Thus, we have isolated the problem to just one server.

The solution:

It turns out that was NTLBis bound to complus1, it was bound to each NIC (both members of "Team"). Once we remove it converges on the NIC team and NTLB nic1 and nic2 the servers in the cluster without a problem.

Summary:

When clustering, you need only the NTLB service is used in one of the network select a cluster. No other network card should NTLB bound to them. As we continue with our proposed configuration, everything works well. All the tests are successful, and it looks like they have the willlarge fault-tolerant production environment. Next is the SQL cluster to live the implementation of two SQL cluster on an EMC SAN, with no tolerance for downtime, this should be fun. Until then ...

Friends Link : customerlogistics counterstrikededicatedservers searchdomainname contact marketing internet marketing advertising kids christmas present