> there were many machines getting improper credit due to database errors. I
> recalculated all credits yesterday so that some may have lost credits, and
> some may have gained depending on the past database problems. sorry for the
> confusion.
>
I've also lost about 13K of credit - seriously not happy.

> there were many machines getting improper credit due to database errors. I
> recalculated all credits yesterday so that some may have lost credits, and
> some may have gained depending on the past database problems. sorry for the
> confusion.
>

That must have been some error! Prior to your recalculation I had approximately 29,000 Climate prediction credits, now I have 945. As far as I am aware I have obtained those credits fairly!

I joined Climate Prediction in September 2004 and had completed 2 full runs plus some odds and ends.

In effect your "adjustment" has in effect wiped out aproximately 10 months of computing time donated by me to the project.

I feel that I should be credited for legitmate work done.

I trust that this situation will be rectified and that people that have lost out will be credited for their efforts.

In my opinion statements such as "sorry for the confusion" do not address this problem.

I hope that these legitimate missing credits will be restored to me.

Please bear in mind that there are other BOINC projects available now.

Hi Carl
well i have this problem too
i just see that i lost about 98 000 credits
most of them are of my home computer /p4@2.8 with HT enabled, 24/7 on and online/
and yes i was use merge few times /i cant remember how much exactly/ and now this computer /id 190533/ have only 7 277.85! credits for all the time /27.08.2004 till now/
how i can get my credits back? :)

> > Michael;
> > 1.how many credits did you lose or gain?
>
> None, as far as I can tell.
>
> > 2.Have you ever used the merge computers facilty?
>
> Yes.
>
> > 3.As far as I am concerned, ALL my credits were correctly gained!!! I DID
> do
> > the work for the 138,000 credits and it has now been taken off me.
>
> Then tell them about the results that are missing. I have archived every
> result I've done, including back to pre-BOINC days.
>
> Michael

Well congratulations to you for having your stuff saved... How ridiculous! We shouldn't need to save everything forever to show what we have EARNED when someone screws up the database/scores when they FIX (?!?!?) something! I shouldn't even need to worry about someone screwing up my credit at all. Give me a break!

Anyway, for the more logical side of this thread, I also am missing almost a third of my EARNED credit now! I had 13232.45, but now it says I have 8979.16, so I am missing 4253.29. It appears as if I have lost all the credit I had EARNED on 2 of my computers before I had to reload them & then merge them. (I know this is nowhere near the numbers that some others are missing, but it is important to little old me, all the same!)

I would greatly appreciate it if somebody could please fix this for me! If it's not too much to ask, it might be nice if we didn't have to worry about this kind of problem being dropped on us. I suppose if that is too much to ask, then I guess there are much more considerate people running other projects that I could better spend my processing time with!!!

Thanks, McPDR

The credits have been recalculated. Just to clarify, if your results took a big drop recently (most likely before the final recalculation which was just completed), it should be back to normal now that I have sorted things out using the trickle information. This information happily/luckily has the correct hostid/userid info that the database problems & merge messed up when the database was in a bad state around June.

The lost credits you may have experienced temporarily had nothing to do with CPDN trying to arbitrarily "reinvent" how we do credits, or claiming that people cheated, or anything bad. It was just part of the "disaster recovery" as I built up the result host/user id fields from the (more reliable) trickle data. I have added a new function in the merge to backup the original host id to further help prevent this in the future. Hopefully the database is now fine-tuned and this won't happen again.

> The credits have been recalculated. Just to clarify, if your results took a
> big drop recently (most likely before the final recalculation which was just
> completed), it should be back to normal now that I have sorted things out
> using the trickle information. This information happily/luckily has the
> correct hostid/userid info that the database problems & merge messed up
> when the database was in a bad state around June.
>
> The lost credits you may have experienced temporarily had nothing to do with
> CPDN trying to arbitrarily "reinvent" how we do credits, or claiming that
> people cheated, or anything bad. It was just part of the "disaster recovery"
> as I built up the result host/user id fields from the (more reliable) trickle
> data. I have added a new function in the merge to backup the original host id
> to further help prevent this in the future. Hopefully the database is now
> fine-tuned and this won't happen again.
>
>

Thanks Carl!
I'm back on board with the project I am passionate about.
This is really important stuff that I am dedicated to.
I don't want credits I am not entiltled to but I do want recognition for work done.
Your loyal suppporter

> This information happily/luckily has the
> correct hostid/userid info that the database problems & merge messed up
> when the database was in a bad state around June.

Hello Carl,

It seems I got my points back, thank you!

Still all's not well in camp, though. All my credits from "old merged machines" have been attributed to a single computer (lpi), which certainly hasn't done this staggering amount of work...

Not that it really matters (I've got the correct amount of credits), I am just curious though, as to how the credits ended up with this single machine, as the correct hostid/userid info should still be available?

Please enlighten me whenever you find time...

Hi, since the old machine is deleted by boinc when you merge, I just had any credits pointing to the old (deleted) machine go under the latest machine id (just to havfe somewhere to put them).

> Hi, since the old machine is deleted by boinc when you merge, I just had any
> credits pointing to the old (deleted) machine go under the latest machine id
> (just to havfe somewhere to put them).

Well, I should have guessed that, shouldn't I!

What puzzled me was the fact that 2 or 3 days ago these points were actually assigned to other (still existing) machines.

But I see your problem (now) having to start rebuilding the points from scratch. It would probably be technically possible but not physically possible to reassign the points to where they were. And certainly not worth the effort...

Thanks for your reply!

Thank you Carl!
everything looks fine now :)

> The credits have been recalculated. Just to clarify, if your results took a
> big drop recently (most likely before the final recalculation which was just
> completed), it should be back to normal now that I have sorted things out
> using the trickle information. This information happily/luckily has the
> correct hostid/userid info that the database problems & merge messed up
> when the database was in a bad state around June.
>
> The lost credits you may have experienced temporarily had nothing to do with
> CPDN trying to arbitrarily "reinvent" how we do credits, or claiming that
> people cheated, or anything bad. It was just part of the "disaster recovery"
> as I built up the result host/user id fields from the (more reliable) trickle
> data. I have added a new function in the merge to backup the original host id
> to further help prevent this in the future. Hopefully the database is now
> fine-tuned and this won't happen again.
>
>

Hi Carl,
I have a problem with earned points, and also with lost work unit (at least from list of my workunits. Currently I'm working on result 2l0o_300141771_0, but from 29th July I can't see it in list of my result anymore. Also my points drop from 7562 to 755. Affected HostID is 166724, and the result is still in progress (currently in 57,50% of calculation). I never merge computers. Can you please chcek it?
Thank you.

You lost 6805 credit which is the amount for a complete WU. Do you have a resultid reference? From what you have said, it seems to me possible that someone else completed it and the credit was reallocated to them because they got further with that WU.

You may want to suspend itand work on a different WU, in case Carl works out a way to save and use it.
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Climateprediction_FAQ">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

&gt; You lost 6805 credit which is the amount for a complete WU. Do you have a
&gt; resultid reference? From what you have said, it seems to me possible that
&gt; someone else completed it and the credit was reallocated to them because they
&gt; got further with that WU.
&gt;
&gt; You may want to suspend itand work on a different WU, in case Carl works out a
&gt; way to save and use it.
&gt; _______________________________
&gt; Visit <a> href="http://boinc-doc.net/boinc-wiki/index.php?title=Climateprediction_FAQ"&gt;BOINC
&gt; WIKI</a> for help
&gt;
&gt; And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the
&gt; news in one place.
&gt;

Thank for reply.
I don't have a result ID, as I wrote, I can't find it in my list of results. So it really seems as it was transfered to someone else (or possible someone finished it before me). But is it so, that each workunit is send just to one host and it waits for return (or failure)? Or is it each workunit sended to more people, and the result is compared between them? If just to one people, why it was removed from my list? I don't understand it and I'm really confused.

Until yesterday, each result id should only have been sent to 1 person. If they want to do cross checks they sent out different resultids which are the same WUid.

Unfortunately there were not only problems with the database running slow and timing out, there were also bugs in BOINC code that didn't catch the failure. Both issues have been fixed so this shouldn't happen again. The consequence was that quite a lot of resultid were sent out twice instead of once. About 14 of 15 were probably failing and mostly very early in the run. Either computer can return the result. So only around 1 in 225 of the resultids sent out twice would result in more than 1 person getting very far with the same resultid. If there is 1000 cases of reallocated credit, there is probably less than 100 people like you where there is significant unnecessary duplicated work.

In an ideal world you would have been emailed to tell you to abort it earlier. Unfortunately your trickles probably weren't recorded as the database is set up to prevent information being overwritten. With no record of where the problem was affecting people, it would involve writing more code to try to trap the attempted trickles to get the information necessary to email people. The project scientist have other work to do, not enough time, and too few people were affected to make it that much of a priority. Easy way to spot the people - remove the credit so they complain here.

It isn't wonderful that it happened, but a lot has been done to deal with the problems arising.

(Since yesterday some resultids are being recycled where there have been no trickles and the WU is nearly a year old.)
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Climateprediction_FAQ">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

&gt; Until yesterday, each result id should only have been sent to 1 person. If
&gt; they want to do cross checks they sent out different resultids which are the
&gt; same WUid.
&gt;
&gt; Unfortunately there were not only problems with the database running slow and
&gt; timing out, there were also bugs in BOINC code that didn't catch the failure.
&gt; Both issues have been fixed so this shouldn't happen again. The consequence
&gt; was that quite a lot of resultid were sent out twice instead of once. About 14
&gt; of 15 were probably failing and mostly very early in the run. Either computer
&gt; can return the result. So only around 1 in 225 of the resultids sent out twice
&gt; would result in more than 1 person getting very far with the same resultid. If
&gt; there is 1000 cases of reallocated credit, there is probably less than 100
&gt; people like you where there is significant unnecessary duplicated work.
&gt;
&gt; In an ideal world you would have been emailed to tell you to abort it earlier.
&gt; Unfortunately your trickles probably weren't recorded as the database is set
&gt; up to prevent information being overwritten. With no record of where the
&gt; problem was affecting people, it would involve writing more code to try to
&gt; trap the attempted trickles to get the information necessary to email people.
&gt; The project scientist have other work to do, not enough time, and too few
&gt; people were affected to make it that much of a priority. Easy way to spot the
&gt; people - remove the credit so they complain here.
&gt;
&gt; It isn't wonderful that it happened, but a lot has been done to deal with the
&gt; problems arising.
&gt;
&gt; (Since yesterday some resultids are being recycled where there have been no
&gt; trickles and the WU is nearly a year old.)
&gt; _______________________________
&gt; Visit <a> href="http://boinc-doc.net/boinc-wiki/index.php?title=Climateprediction_FAQ"&gt;BOINC
&gt; WIKI</a> for help
&gt;
&gt; And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the
&gt; news in one place.
&gt;


Thanks for explanation.
It means I should reset project (get a new WU)? The more than 350 hours of CPU is lost. It is not good, but I can live with it without problems.
Is it good to update BOINC to newest version (as you mention some bugs ion BOINC code)? Or do you know which version of BOINC under WinXP should works without problem?
Thank you in advance.

that's strange, but this result was already completed by someone else, although the upload codes are wrong. I guess it's from the period of the database problems though.

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=834807

&gt; that's strange, but this result was already completed by someone else,
&gt; although the upload codes are wrong. I guess it's from the period of the
&gt; database problems though.
&gt;
&gt; http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=834807
&gt;

Do you recommend to reset a slot? I think it is not really important to run it again (It is in about 58%)...

The bugs in the BOINC code that I mentioned were in the server code.

BOINC CC v4.19 4.25 or 4.45 are generally considered the most stable versions for windows.

If you want, you could back up your BOINC directory in case Carl comes up with a way to make your 58% done model usable. Then reset climateprediction.net. (there is no other running model on that host is there?)


BTW Carl I don't see what is strange. It is exactly the sort of thing I would expect to see (not huge numbers but just a few) and if they are going to be noticed at all now would be the time due to change of credit.
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Climateprediction_FAQ">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

&gt; If you want, you could back up your BOINC directory in case Carl comes up with
&gt; a way to make your 58% done model usable. Then reset climateprediction.net.
&gt; (there is no other running model on that host is there?)

Another model is just on another host. One model per host.
I will backup it and then reset. Thank you.