Folding Forum

I have experienced this before with the same WU on different machines, but not since I folded with GPUs. The picture below speaks for itself, but what should I do? Delete the last that come down and start it again with the new machine ID?

The error is certainly not with me since both machines that are involved has gone flawlessly without stop for several weeks.

Picture:

You have a choice. Let them both fold, and get points for each one. Or delete one (doesn't matter which one, but if you do, I recomment deleting the one least completed ) and change the ID so that it moves on to a new work unit.

However, if you delete one WU, it will get reassigned to someone else to be completed. No optimal solution available.

Thanks.
Just out of curiosity! What is the reason why this happens sometimes? If given points for both then it is probably easiest just to let them go as normal, or not?

Several people have suggested possible causes for this situation but I don't believe there ever has been an authoritative answer.

One theory that seems somewhat plausable to me is that when a sever detects a WU being dumped it issues two new copies to increase the chances of it being completed promptly.

That may be what 7im is talking about when he says there's no optimal solution available.

Of course the server has no way of knowing whether it was intentionally dumped or one of those errors occurred which causes the client to dump the WU.

Thank you, I ask because I have experienced getting the message that "this WU has already been received" and then you get no points credited. But it is possible I do not remember correctly here.

When you receive two WUs with the same PRCG (at different times) they might be an identical WU that's being reassigned to the same client or they might be a new copy generated by a server being issued to another client in an attempt to compensate for a dumped WU. You can't get credit for more than one copy of a reassigned WU. You can get credit for a new copy. It's not always easy to understand the difference since to the casual observer, they look the same.

Thank you, so this can be pretty random? I have not deleted any WU from some of my machines, but will the WU be rejected anyway, even if it was originally deleted on another donor's machine?

I remember that this happened quite often when I folded with 24 units of the GPU back in 2010, and I left them usually to get done.

I have not observed it to happen very often but then I don't run 24 identical clients so if the WU was duplicated because of a problem on somebody elses machine, I would never notice it.

In other words, there is no control of the problem, and we just have to live with it?

No, there IS control to the problem. The extra work units are ONLY sent out in response to other errors. It doesn't happen that often, so clearly there is a level of control already in place, otherwise it would happen much more frequently.

And if one specific work unit has a lot of errors, and we see many many copies going out, a Mod or Admin can report the WU as bad, and get it taken out of circulation, which is another level of control.

I understand, but but does it work something like this! In the first hand things happen automatically as a response from the server, and if it happens often, and someone report the error, Admin can make the decision that this is a bad WU, and take it out. Does it work something like this?

Edit: I did get the credit for this WU as normal.

-alias- wrote:...In the first hand things happen automatically as a response from the server, and if it happens often, and someone report the error, Admin can make the decision that this is a bad WU, and take it out. Does it work something like this?...

Nearly. The Server usually have a limit to the number of times the same WU can be reassigned. Under some conditions, it doesn't happen and is reported in this Forum. Admins/Mods may do a look-up on the WU and normally, we report it if there are 3 or more errors, and hopefully it will eventually be stopped from being (re)assigned.