Sorry, no optimization tips here, only a way to monitor all clients on all nodes. The key here is to monitor the value of 'ncfft' in the state.txt file. The wu is complete when ncfft=9024.
Create different que's for each node. e.i. seti1$batch, seti2$batch, etc. The command "sho que seti*" will only show if the que is running, but NOT show if the client is running or stalled. To check on all clients, log in to any node as 'system', and issue the following commands:
The last command will return a value from each client, such as ncfft=7034. After a few moments re-issue the command to be sure the value is changing. If ncfft=9024 the client is either trying/waiting to connect or stalled. If the value does not change after a few minutes, the client is stalled. If so, stop the que, re-issue the startup command with -interactive. After a successful download and the client starts crunching, type Cntrl-C, Then issue the startup command with -submit.