SAH not responding and getting overloaded with predictor work units

Well as the title says this problem has happened over a progression of about three days. As sah is currently not responding it keeps trying to talk to the scheduler what fails and then every so often it randomly decides to download work units from predictor. I usualy have had about 4-5 work units cached but now i have 13 predictor work units down what is about 3-4 days worth of work. Where my work settings are configured to 0.3 to 0.4 days of work. Why is Boinc doing this?

BOINC is becomming utterly convinced that you need more work for some reason. I am having the same problem on two of my machines. I have turned of network access on one of them, but I am letting the other one continue on to see what happens.

> BOINC is becomming utterly convinced that you need more work for some reason.
> I am having the same problem on two of my machines. I have turned of network
> access on one of them, but I am letting the other one continue on to see what
> happens.

Ok thanks for your quick reply. Ill disable network access for the time being as if it keeps adding them by the end of tomorow ill have over 20 work units