Determining Logical Hostname

Question

We have a (non-cluster-aware) client-server application that has several background client programs that do asynchronous processing. In principle, these can be distributed among an arbitrary number of hosts, although in practice usually all reside on a single host.
These clients are administered by a control program. Details of all background clients are kept on a table, which includes the host name. The client program compares that host name with the results of the UNIX program 'hostname'. If they are the same, it controls the program directly. If not, it uses rsh to log in to the correct host.
The problem is that 'hostname' prints the physical host. When failover occurs, this changes, and the table becomes out of date.
Just changing the table contents to the logical host name is not possible, as then all operations would be via 'rsh', and this is generally disabled in production systems for security reasons.
My question is, how can I determine the logical hostname, so that the control program can determine whether an rsh is necessary in a clustered environment?
Thanks,
Roger

Answer 1

Know this reply is a bit late but I thought I'd post it as I'd like to know if anyone has a better way of doing this within a script.
How we achieved it was as follows:
Use uname to determine the physical host name
Use a case statement to define the logical hosts which maybe present on each physical host - e.g:
case $physhost in
physhost1) logihost1=fred
logihost2=albert
physhost2) logihost1=albert etc
The reason for using the case is that we list all logical hosts for all our physical hosts in one script and use the same script on all servers.
Use vxdg list and grep for the disk groups which belong to the logical hosts - e.g:
disk_dg=`vxdg list | egrep -v 'NAME|rootdg|sc_dg' | grep -v ${logihost2}dg | wc -l`
If this command is run on physhost1 and sets disk_dg to 1 then logihost1 (fred) is present on this machine.
Does anyone have a slicker way of doing this?