> How can I reduce this figure to a reasonable number.
You don't. If I remember correctly, someone was trying to balance the estimated time on one of the platforms, and goofed. Not much you can do about it, except to continue to crunch through the affected work units until they are gone (not every work unit is affected).
> For my P4 it is set at 48 hours and for my P2 it is 113 hours.
Yes, I see similar times as well (28 hours on my XP1800, > 100 hours on my P2).
> With numbers like these, I have to set my cache to 30 days just
> to get 3-4 days ahead. These machines run 24/7.
Please don't set your cache that high. Seti@home work units 'last' only 14 days until they expire, and the scheduler won't (shouldn't) give you anything it thinks you can't finish in time.
In my opinion, if, as Jaaku implies, Berkeley is using this 'oops' to help ration out scarce work units, you should not be trying to 'get greedy' anyway. Just leave your cache at the value you mentioned (3 to 4 days), and accept the 'dry times'. If you, and a large number of other people, keep trying to 'game the system' into giving them a larger amount of work than it wishes, the 'dry times' are lengthened for everyone, and Berk will have a lot harder time trying to build up a decent surplus of work. If you 'game' the scheduler into giving you 6x the number of work units it would normally give you, thats 5 other people you have screwed into sitting dry.
So, please don't run up you cache preferences to some rediculous amount. Please just leave your cache preferences at some normal value (the figure you mentioned of 3 to 4 days is good, in my opinion), and be patient on Berkeley to work things out.