I see that almost every day on most of my machines. It's normal. For example, the OPN1 tasks aren't all created equal in size/length, so some of mine finish in 1.5 hours vs. 2.5 hours on the same machine. The BOINC scheduler does some calculations based off a running average (I think) as well as other factors such as benchmarks, buffer size, and how long the machine runs/day and then adjusts accordingly. So if my ETA goes from 1.5 hours for my 1 day buffer, but then the last few work units take longer causing the buffer to adjust to 2.5 hours, the job cache becomes "full" for a 1 day buffer, so the scheduler doesn't need to request work.

tl;dr: It's normal behavior even on machines that run 24/7 with no other changes. The only "change" is the running average of the tasks.

Oof, the math gets trickier when dealing with download limits on the server-side. In my opinion, a 10-day cache is far too long, since 10 days is the reporting deadline for HSTB, ARP1 (I believe) and other projects. OPN1 and MCM1 are 7 days. I'd lower that number, personally. Even 2-3 days is generous. I'm comfortable with 1-2 day buffers should some thief steal Comcast's cable junction boxes in the middle of the night causing a long outage. The only time I hoard is if a project is coming to an end, and I'm very close to a badge goal. :P

Sorry couldn't help further; the maths is a bit too much for my head at 1 AM. I wouldn't worry so much about that message to be honest.