I'm almost completely shutdown as far as NPLB goes. I have no more than 10 cores hitting Carlos' Server right now (five of them from work).

From my llrnet proxy logs, router logs and direct observations the problem was two fold. First, Carlos's proxy could not keep up with the demand for connections and his pruning process which removes completed tests from the knpairs and joblist files was getting overwhelmed.

* I had an almost constant connection from the home farm when his proxy was up. But everytime his proxy was up my proxy was flooding his with results (2000/hour). I tryed shutting off access to his server to allow others to connect for a while, but that just increased the load on his proxy once I allow connections again. (At one point I had over 5000 results waiting for upload) And caused his proxy to shutdown quicker.


Second, my own proxy was getting overwhelmed with connections once I got above about 40 cores, by the time I got 60 cores it would completely lock up.

* I tryed setting up a second proxy to run from Carlos' port 300 proxy, but it locked it up too, so turned it off. As a side note, once I reached 60 cores the traffic bogged down my own network, I could not reach my webserver or dnet proxy and websurfing slowed to a crawl.


I really like the project so will be trying something different to get going again, I think moving the proxies over to the Linux boxes will be first step and perhaps setting up 2 seperate subnets instead of everything running off one. Problem is getting the time to do all of this.