Symetrik Domino Consulting

We have two hub servers in our environment that are clustered.

We use these as replication hubs to all of our spokes (local and remote sites). Additionally, these act as sort of a mail hub since all of our remotes sites are not directly connected to our home office site. Lastly, they also serve as outbound SMTP gateways.

Any mail that is sent from Chile to a user in Spain gets routed through Hong Kong since their is not a direct connection from Chile to Spain.

All servers in each site are in their own NNN (Notes Named Network). The Hong Kong site servers are in their own NNN. The hubs are in their own NNN.

This allows us to route mail through Hong Kong where necessary using mail connection documents.

We came up against a problem recently where one of the hub servers couldn’t route mail to a remote site in S. America due to some sort of networking connectivity issue. We also couldn’t route mail to the Internet.

Initially, we had a connection document to hub1 and a connection document to hub2, then we had a connection document to the cluster. Typically, this connection document worked as a failover for either hub if one or the other was down. Most mail routes through hub1, since it is first in the list.

NOTE: You can replicate to a cluster instead of directly to one of the cluster mates. What is required is individual connection documents (with resolveable host name or IP address specificied in the connection document), which can be set as disabled. This lets Domino know how to contact each of the cluster mates. Then create a normal connection document (mail and/or replication) as per normal, but in the server name field, specify the cluster’s name.

When we discovered the routing problems of Hub1, we deleted all of the TO > Cluster connection documents, so that from each spoke, there were no cluster connection documents, only direct connection documents to the cluster members.

(That reminds me, what is the official IBM/Lotus phrase Cluster Mate or Cluster Member? I have seen both in the documentation).

Then enable replication (if necessary) and Mail routing. On the routing cost field, we set Hub1 to 10, and Hub2, we kept at 1.

Once, we made the changes to the NAB for all spokes, we pushed the changes out to all of the servers through a manual replication.

We then issued the command to manually refresh routing tables in the router with this command:

Tell Router Update Config

Note: we are on Domino 6.5.6 FP2 on most of our servers. There is no longer a need to restart the router.
Note2: You can manually route mail that will not route after making configuration or connection document changes by using the “route *” command.

In just a few seconds, all mail was routing through Hub2 instead of Hub1.