Hello vcd683s,
2 GB RAM is plenty for 2 MCM jobs. I assume that you are getting your checkpoint data from BOINC Manager - Tasks - (select work unit) - Properties. What CPU percentage does Task Manager show for your work units?

Lawrence

The average time per WU has been increasing of late. Here is a quick chart of the average time per WU for each date. Since the average time is increasing, there are probably some long units in there.

Cheers

Do the work units that need to get sent out again get sent out to more reliable and/or faster hosts? Would seem to make sense.

I don't think you get the point. Normally I do not have WU's that abort because of passed deadlines. The scheduler is probably too optimistic. This is the first time that this is happening...

Yes, I got the point that the machine was overscheduled, probably due to the variable length nature of the MCM1 project work units. Whatever the cache setting was could have been a contributing factor if the cache was set to more than 4 or 5 days. However I only tried to answer the question which was posed about the reissue of work units. Due to the sometimes highly variable nature of the length of these work units, in my humble opinion, I would not set my cache for more than 2 days. For instance on one of my crunchers, which is solely dedicated to WCG, the work units have varied from only a couple of hours to over 20 hours. If your cache is too large and your machine runs into a batch which takes 5 to 10 times longer than anticipated, you will run over the deadline with some units. However, all that being said, those work units will simply be re-issued to other machines and will get completed on a new schedule. Not to worry, these kind of little blips in the system are easily handled.
Cheers