In the test I did it retried once per minute, not once per second. We are both running 5.10.45 but maybe you have a different build? It might be worth trying the latest version because even if we pinpoint a bug in the source they are not going to release another 5.10.x. Your options will be to compile a fixed 5.10.45 yourself or update to 6.2.18.

BOINC has >100 results in upload queue.It retry sending of each result once per minute. But it retry 2 results at once, slightly later - next pair and so on and so on.
So, each second some of results being send.

And it pretty meaningless, to retry next pair right after prev pair fails. It could check if project site accessible (as it does with reference site) and if project down, stops all retries until project will be reachable again.
It can be applied not only to situation when whole network is down (as in case of cable unplug), but when some of projects are down for maintenance too. FOr example, SETI project down few hours each week.

ADDON: Why such retries are evil? - Because boinc.exe eats much of CPU in this situation. Now, after ~40h of network outage, it consumed 2h19min of CPU time (on Q9450 system (!) ). 4% of CPU time for quad system - it's too much for management tool.

@Ageless. How can I lower CPU consumption of BOINC 5.10.45 in this situation? I didn't change any "reminder" options by hands, should I increase it and where?

---

When host runs with broken network (for example, network cable unplugged), boinc.exe (runned as service) begins to consume CPU (~2h of CPU time for ~10 hours w/o network).
At this situation BOINC client manager can't connect to boinc service and hangs (not just shows message like "Cant connect to core client" but exactly hangs).

There is a bug that causes a hang if network is unavailable; unrelated to the high CPU usage.

---

When host runs with broken network (for example, network cable unplugged), boinc.exe (runned as service) begins to consume CPU (~2h of CPU time for ~10 hours w/o network).
At this situation BOINC client manager can't connect to boinc service and hangs (not just shows message like "Cant connect to core client" but exactly hangs).

There is a bug that causes a hang if network is unavailable; unrelated to the high CPU usage.

Maybe. If so, there are 2 separate problems instead of one :)
I can't reproduce "hang" still, but high CPU consumption is easely reproduceable.

---

it happens to me too... I have 4 cores, and they run at night... and in the morning, screensaver says Boinc is loading and then one pop up says no network availabe and manager hangs and closes... I will try it crunching with 3 cores to see if it works..