You don't tell us what client version you have. The latest development is still having scheduling bugs, so plz stick to 6.10.58 [recommended by WCG] or 6.12.3x

Long as your client is not having idle threads things are fine. Eventually when work IS needed, the client will fetch new tasks and SAME TIME clear the "Ready to Report", which will happen no later than 24 hours after the oldest completed.

--//--

Before hitting update select WCG then properties. There could be a deferral counter running. Else, all is fine. Hitting Update causes unnecessary server scheduler activity, which is why the client saves up the Ready to Reports, so to accumulated reporting only requires the system to whir once for the 15 and not 15 times. The mentioned development client will likely cause even more worry [for the micro-managers] as it's designed to reduce the client / server connects even further along the line of: "Let's first empty that cache down to the lowest "connect every..." interval and then backfill in one big download session."

Remember: You may want to mm, but the outset of BOINC volunteer computing is... Set and Forget and only notify when there's a problem [Notices tab of the 6.12 and up clients] :D

--//--

Depending on which projects you are running, the scheduler can get some interesting ideas about how long the currently downloaded work will last.

For example, assume that each task runs for 2 hours and you have a queue of 24 hours, the scheduler will download 12 tasks. What happens is one of those tasks runs for 3 hours and suddenly BOINC thinks you have too much work for your queue because the 11 left will (theoretically) run for 33 hours.

What it will do in this situation is not report or download any new work for 9 hours when it assumes there is 24 hours left to match your settings. If you extend this logic out to larger queues you can easily see where it might not report or download for long periods.

As Sekerob said, it will report tasks no later than 24 hours after the first (oldest) one finishes. Short version is don't worry about it. If it's a real issue, maybe selecting a lower value for your queue would be the answer.

There's a "connect every" type setting in the device settings page that controls how often the machine will connect to the grid. I wouldn't expect a machine to report work any more often than that setting without outside action. The setting is in days, so you need to make sure it's less than the size of your work cache, but any more often than a couple of times a day doesn't really achieve anything in the long run.

This is probably not the issue but: If any WU is suspended new tasks will NOT be requested after reporting completed tasks.
ie: 1/6/2012 8:50:33 AM | World Community Grid | Reporting 1 completed tasks, not requesting new tasks
Additionally when the data base in inaccessible the project will show a status of "Communications deferred 00:59:00 and count down. The message "1/6/2012 8:48:32 AM | World Community Grid | Server can't open database" will show in the event log and nothing will be reported. Anyway, Just in case...