[Extension of my upper post]




BTW.
My pending credit: 295,397.36

So thanks again for your PN to Eric.

But since I have received no news yet.

Good idea your call here!
Please who can see their credits Pending NOT?
I did the last time you saw a week ago and 550,000 since they were high, then they should be at ~ 600.000 but I can not see her anymore.

---

And every 4 hours the UL server have 3 times probs to receive my ULs.. again 4 hours break.. and again 4 hours break.. + 100 , + 100 , .. results in UL overview..

This is true only if uploading is always a problem, and that has never been the case.

It only seems like there are always problems uploading because no one writes about things going well.

If you load up 6.10.4 (or 6.6.38), see it start the back-off and say "oh, no good!" and immediately go back, you'll never be able to tell if it works or not.

I haven't read the code, but it looks pretty random. I need a bigger sample before I know.

---

And every 4 hours the UL server have 3 times probs to receive my ULs.. again 4 hours break.. and again 4 hours break.. + 100 , + 100 , .. results in UL overview..

This is true only if uploading is always a problem, and that has never been the case.

It only seems like there are always problems uploading because no one writes about things going well.

If you load up 6.10.4 (or 6.6.38), see it start the back-off and say "oh, no good!" and immediately go back, you'll never be able to tell if it works or not.

I haven't read the code, but it looks pretty random. I need a bigger sample before I know.

I have to add one more comment.

Sutaru is looking at what he thinks will happen on his computer when he's one of the relatively few running with the "project back off" code.

We need to look at how this works when most people run it.

Let's say there are thousands of machines trying to upload, and half of the uploads are failing because the uploads are just plain coming too fast.

Half of the BOINC clients out there have errors, and basically disappear for a couple of hours. We'll call them group "A".

The other half are group "B" and they keep trying, and most of the uploads get through because there is less congestion.

When group "A" finish their timeouts (not all at once, of course), they don't have to compete with the timeouts from group "B" -- those uploads finished smoothly.

The size of those two groups aren't important. Failed uploads make group "A" bigger, and group "B" will tend to be whatever can get through smoothly. As the group "B" machines finish, or if server problems get fixed, machines in Group "A" will tend to move into "B".

But it won't work if people dismiss the idea out-of-hand.

---