Severe Credit Penalties due to Server Issues

[rusty]

Hello,

I was just watching the logs for a work unit that upon reaching 100% progress had an Estimated Credit of ~140,000.

I then watched as the client failed to connect to the collection servers several times... then saw a handful of successfully initiated yet ultimately failed upload attempts.

By the time the work unit was finally accepted by the collection server successfully, the reported "Final credit estimate" was 50% less (~70,000).

I assume this is because the collection server assigning credit has no way of knowing that receipt was delayed due to network issues. So, it is assumes that it simply took me much longer to complete the computation than it actually did.

Would it be possible to assign bonus credits or something to help offset this?

I know this is literally at the bottom of the list right now as keeping the system running is clearly paramount, but I feel it should at least be on the radar. Thanks for all your hard work! This project is truly doing something good.

[v00d00]

Unlikely you will get bonus credits. This is not a new situation and has happened in the past. What makes this different is the scale of the problem. Whenever folding has an expansion or releases something new we get an uptake of donors. This time we got a lot more than on those occasions. Add on to that not enough servers, workunits and general infrastruture to handle such a massive increase.

When you run this project you just havve to accept its not perfect and things will go wrong at times. Points at this point are probably not even under consideration due to the worldwide pandemic that will probably wipe out a chunk of the worlds population.

[JimboPalmer]

The bonus name is honest. It is a Quick Return Bonus, not a quick completion bonus. Sorry

Adjustments ARE made if the bonus calculation breaks, as has happened. We all got base credits, then months later they fixed it, and we got the QRB bonus then. (and other volunteers complained as we got points but showed no WUs. They thought we were cheating)

[rusty]

Thanks for the replies.

I am new the Folding@home project, so I did not know if this issue was already known or in some way being mitigated on the backend.

I don't want to be misunderstood here. Clearly, points don't matter. We are all contributing resources to solve very real problems in times of need. Higher point totals don't solve problems. Returned work units solve problems.

I was merely making sure this issue with the accounting system was known and was wondering if it was being somehow offset.

I noticed this issue because I was retargeting a GPU cluster to process COVID-19 work units. At the time, I was monitoring point accrual on folding.extremeoverclocking.com to make sure everything was running smoothly. When I saw dips, I went to check individual node logs and found what I reported in the initial post. Clearly, using points as a surrogate measure for processing throughput was a flawed idea (on my part) since points also bundle in network connectivity issues to a far greater extent than I expected.

[Nathan_P]

Thanks for the replies.


I noticed this issue because I was retargeting a GPU cluster to process COVID-19 work units. At the time, I was monitoring point accrual on folding.extremeoverclocking.com to make sure everything was running smoothly. When I saw dips, I went to check individual node logs and found what I reported in the initial post. Clearly, using points as a surrogate measure for processing throughput was a flawed idea (on my part) since points also bundle in network connectivity issues to a far greater extent than I expected.

i usually do something similar when folding is running under normal conditions. if my points drop I tend to go looking for the issue, otherwise the 2 rigs just sit there doing their thing. Given current conditions I would say look at your WU completion totals. if that looks normal then all is good.

Not sure what you are using for stats but I use :- https://folding.extremeoverclocking.com ... ist.php?s=