My connect is set to .01 days. I'll shorten from 8 days back down to 1.5 day cache after I get my Sapphire in Flu. I really didn't want any Beta this round due to Flu ending so soon.

So I'm not upset about not getting any this round.

I have been told this before. It means that with a 5 day buffer and a 4 day deadline, any user could download a Beta job, which would then be run in high priority mode so that it will complete before the 4 day deadline. I have been unable to confirm that this is how it works, but that does not mean this is incorrect.

According to the techs there is currently no preliminary screening of devices based on their speed, usual turnaround time, "reliability" or whatsoever.
Therefore, when a device sends a request to fetch work and beta WUs are ready for distribution, if this device is participating to the beta program then the normal scheduling rules apply for sending it a beta WU or not.

Among these rules, from other past messages it appears that the "Connect every..." parameter is critical, not the "Cache extra work" one. I am not too sure about the "Switch applications" parameter but I would assume that if it were not OK the client would not send a request to the WCG project at that time.

Once the WU has been received by the client it will be handled the same as a rush/repair WU: if these WUs are usually raised to high priority immediately the beta will be too. If it is only when the deadline is closer it will be the same for the beta WU since they have the same 4-day deadline. This behavior is depending on the "Connect every..." and on the "Switch applications" parameters, even if this device is crunching only for WCG.

jasm580's comment is particularly relevant: devices runniing with a large cache are more exposed to long periods without any request for work because if the completion of a longer job causes all estimated durations of the jobs in queue to be increased significantly that may result in the sum of all these durations jumping up by many hours during which the client will not need to contact the server. This effect is minimum on devices running with a short queue.

I hope this helps. Jean.

OK now I get it. Connecting often is critical but if your cache is full it will not request work. Not sure if a small catche makes a difference as ones computer only requests new work, all things being equal, when work is completed to mantain the cache full to one set level. That is if one has selected a 2 day catch or a 8 day cache and they are full and has 2 WU's processing at the same time until one is WU is finished no matter how often ones connects no new work units will be down loaded. Otherwise it would overfill the cache you selected. This is what I understand. Thanks for information.

Thanks for clearing that up for me. I was under (the mistaken) impression that cache came into play also.