automatically starting F@H clients on network

[coral.khl]

hey all,

We're busy with a project for school. We have a Debian server with Nagios monitoring application. We are monitoring 20 PC's (running Ubuntu) for their CPU load, memory load, hdd status, etc.
We want to contribute to F@H community and start F@H clients on those 20 PC's we are monitoring, remotely, depending on their CPU load. We also would like to monitor the status of the folding clients from our server.
Is this possible? Would anyone know any existing projects or applications doing something similar to this? If not, any ideas to put us in the right direction?

Any input will be deeply appreciated and help us contribute to F@H.

thx!!

[7im]

Hello coral.khl, welcome to the folding forum.

There are many ways to accomplish your goals. For instance, there is a monitoring application called HFM that works well on linux to monitor clients across a network.

Let's back up a few steps first. Are the systems single core, or multi-core? And will fah be running full time, mostly full time, or only part time? The answers will determine the best client to run, and that may influence how you want to start up those clients...

[coral.khl]

hey 7im, thx very much for your quick response.

The systems have multi-core CPU's and we would like to run F@H clients as much as possible, but as long as this doesn't effect the PC's performance while user is busy working with it.
We first thought of starting F@H clients (with a low CPU Usage Percentage) whenever the PC starts or the user logs in but in this project we want to accomplish doing that depending on the CPU load (say 5mins after login), as we read it with Nagios from our server.

I appreciate your input and I will check out HFM meanwhile.

[bruce]

Traditionally, FAH is designed to run at an extremely low priority so that the performance of foreground activities is not degraded. For the SMP or Uniprocessor client, that's a pretty good assumption and no monitoring of the foreground CPU load is required. (It's less true for the GPU client, but I don't think you're talking about that one.)

How much free RAM do the computers have when somebody is using them? As long as that's not a limitation, I doubt that anybody will notice the difference unless they look at see the CPU pegged at 100% or they hear more fan activity. Paging delays can be noticed; delays associated with a foreground task preempting a background task cannot.