Long WUs

I'm sorry, but there is no such thing as "long WUs". There are different science apps which may take different times, but for any one science app, the size of the tasks are pretty similar. With older architecture like your P4, the crunch times are going to be rather slow. The other point to realise is that Einstein doesn't send anything unless BOINC requests, so the real question is, "Why does BOINC request so much work?"

I notice that you received a whole bunch of work on 24th May in one big hit. Prior to that, you seem to have last requested work on 16th May. I also notice you have two apparently identical hosts listed, one running BOINC 6.12.43 which last contacted on May 7th, and one running BOINC 7.2.42 which did the big download on May 24th and is having trouble now. I assume these are the same physical machine and that you recently upgraded from 6.12.13 to 7.2.42, inadvertently getting a new host ID in the process. Your current machine shows as having been created on May 12th.

If these assumptions are correct, the most likely cause of your problems is the work cache computing preference settings which you should check and change if needed. If you check the Network Usage section of your computing prefs, the first two you find are the ones that control the amount of work you download. With the advent of BOINC V7, the meaning of these two has changed. This is something the BOINC Devs did at the time when V7 was released.

The first setting (Computer is connected ...) controls the 'low water mark', ie, the minimum amount that will be maintained at all times. The second setting (Maintain enough work for an additional ...) controls how much extra on top of the minimum will be asked for. So unless you want your cache of work to jump up and down through a large range, you should set the first number to be your required cache size (say 5 days) and the second number to a very low number (say 0 days). This would give you 5 days of work at all times with regular top ups as required to maintain that number. If you have these settings the other way around (I suspect you may have something like 0/10), your host won't ask for any work until the cache is empty (0 days) and will then ask for a big hit (10 days) all at once. It would be the last ones of this big hit that are now having deadline problems.

Einstein cannot hog anything. BOINC is in control and is just trying to follow your wishes.

Unfortunately, only you can fix this by adjusting your prefs. If the above analysis of the situation is not correct, please respond with the details of the work cache settings you are actually using.