A win with Lotus

A little while ago I wrote a couple of comments about a problem we were having with cluster.ncf.
After messing around with Lotus Support for ages, it turns out the be a known limitation, but only "Lotus Internal", not publically available.
When I closed the call this morning, the Analyst told me that the solution is now a public document - here.
Hopefully this will help someone else avoid this problem in future. Here it is.

Technote 185700:
Solution
In this particular case, the problem was due to the number of clusters in the customer's domain. By default, the CLUSTER.NCF text file is limited and can reference only forty (40) clusters (assuming only two servers per cluster). Any additional clusters added in this domain will overwrite existing entries in the CLUSTER.NCF, preventing the next hop from being found because the previous cluster/server information is no longer available. Because the CLUSTER.NCF is a log of information cached in memory, and is updated only when the server is shutdown, it will be necessary to restart the server to view the most current information.

To workaround this issue, implement the following NOTES.INI parameter for all servers in the domain:

Cluster_Name_Cache_Size=100

This configuration change will need to be replicated to all servers. The servers involved will need to be recycled. The Cluster_Name_Cache_Size=100 NOTES.INI parameter with the assigned value will allow more than the default of 40 clusters, but sets a new limit to 100 clusters.