CE not counting points correctly

Dennis Nawrot ID: 6360 Posts: 32
21 Sep 2014 11:28 AM

CE Accountmanager has in my case created two BOINC Accounts with different CPIDs. Unfortunately, the main part of my production within these prize draw period was produced on the account which was NOT counted for my prize draw entries.

I'm pointing to that because most of us are running CE/CE BOINC machines without giving much attention to it. If i wouldn't have been so astonished about my few entries in this prize draw, i wouldn't have taken a deeper look on it.

My boincaccount, given by Charity Engine, is CE6842. You can find both CE6842 here:


Which explains to me why ~850000 credits in PrimeGrid are only ~5000 total points worth - they weren't counted.

Mark McA ID: 179 Posts: 220
21 Sep 2014 11:46 PM

Hi Dennis,

Thanks for the report - that's a new one. We'll take a look and get back to you asap.



DocH ID: 526372 Posts: 77
22 Sep 2014 04:00 AM

As far as My entries are concerned I see no problems!

However the following IS very curious....RE "multiple accounts", I have noticed when in the Managing Accounts area of Win 8.1 Pro, I see the following Acccounts:


Local Account



Local Account

I THOUGHT I had rid the PC of all but  The Charity Engine Manager/accounts for managing and running boinc stuff.

I hope I said the above correctly!!

Ideas or thoughts or corrections for Me?

Maybe after The Drawing, If there is indeed a problem or a better way to have my accounts set-up,The Expert/Manager here @ Charity Engine can show Me how to have it All set up right.




Dennis Nawrot ID: 6360 Posts: 32
22 Sep 2014 11:36 AM

Hello DocH,

as far as i can tell, these accounts are needed by the boinc client. I don't use Windows very much, but i can find these local accounts on every windows machine where BOINC is installed.

My problem wasn't based on Windows Accounts.

Every time you register with a new boinc project, this project gives you a "number" (the CPID). It's the real key to your account on the specific boinc project. If you've got already such an CPID, given by other projects you started earlier, you'll have two different CPIDs for a short time.

That's because your new project isn't aware of your already existing boinc identity. Normally, a BOINC-Client where you are running both projects (the new and one of the ones which you have joined earlier), will fix that in a few days, telling your new project that you have already the CPID xyz instead of CPID newxyz, the project corrects itself, and you are correctly registered in every of your project as CPID xyz.

In my case, this didn't happened (until today), and CE checked only the amount of work from my old CPID, ignoring the work of the new CPID. Unfortunately, this was the CPID with the main part of my production within this period.

However, the CPIDs have merged today and CE took over at least my average production. This leads to the funny thing, that i have produced 5228 points in the last months, with a daily average of 8727.


DocH ID: 526372 Posts: 77
22 Sep 2014 10:52 PM

Thanks Dennis N!

Really appreciate the information.



Tristan Olive ID: 22 Posts: 373
23 Sep 2014 07:27 PM

Hi Dennis, did you have any projects running in your BOINC client before you attached to the CE account manager? It's possible for conflicts to occur in that case, which could have blocked your CPID from merging.

Also, did you change anything the past few days that could have resulted in your CPID merging yesterday? Were any projects added or removed? Merging takes place at the client, so most likely something changed there to fix it.

Dennis Nawrot ID: 6360 Posts: 32
23 Sep 2014 08:51 PM

Hi Tristan,

Actually i'm watching my new divided ce6842 accounts (yes, again divided after merging) and i came at least to a hypothesis.

I'm using two machines for CE. The first one runs CE besides some other projects afair and i let Acc Manager do what he wants. The second machine runs only CE, but with rosetta manually disabled.

I didn't change anything in the past few days, but i think something happened like that:

I didn't know that it was even possible, but actually it seems to me as if i have configured some kind of CPID Ping Pong. PC 1 merges, PC 2 claims to have the "right" CPID and divides again. CE seems to check only the first CPID and calculates the points.

I always thought that a merging of CPIDs would be a final step, without any new manipulations by client side. Guess, i was wrong.

I will allow rosetta to run on PC 2 to merge th CE6842s again for a few days, than i will detach PC2. It will be interesting so see if the credits are lost or will reoccur in the next period.

I apologize for the inconvenience.

Correct me if i'm wrong, but i think the thread would be better called "what can happen if you manually configure CE account manager while running on different machines".



Tristan Olive ID: 22 Posts: 373
24 Sep 2014 05:08 AM

I can't speak to the CPID "dividing" again after merging. As you say, that should not happen... but I can say that Rosetta being disabled on the second machine likely is preventing the merge from taking place at PrimeGrid. Since the PrimeGrid account is kind of sequestered off on that second machine all by itself, it can't compare timestamps with your other project accounts and merge to the oldest CPID.

If you're willing to enable Rosetta on PC 2, at least temporarily, I think that will do it.

Dennis Nawrot ID: 6360 Posts: 32
24 Sep 2014 06:14 AM

Sure, but shouldn't have "first machine" have done all the merging stuff? Even if second machine never gets this information, it should affect only machine 2.

However, it was my fault and not CE's. I guess i should move CE on my gerasim-at-home-windows-virtual-machines and crunch ce units instead of bakerlabstuff or prime numbers.

Let me repeat the bottom line for all quick readers:

CE counts correct.

Tristan Olive ID: 22 Posts: 373
24 Sep 2014 02:01 PM

From what I can tell, PrimeGrid is the only project that has a different CPID. Are you saying it is running on both of the machines, not just on machine 2? If only on machine 2, that is by design. Different projects are assigned to different machines under some circumstances. So the only thing that should be preventing it from behaving properly is that Rosetta was disabled manually.

(And actually now that you've enabled that again it looks like your PrimeGrid CPID has merged, though boincstats hasn't yet picked up the change.)

Dennis Nawrot ID: 6360 Posts: 32
24 Sep 2014 03:23 PM

Hey Tristan,

of course it is running on both machines. As i said, i left machine 1 untouched and disabled rosetta only on machine 2. CE Account Manager has run primegrid on both machines.


Tristan Olive ID: 22 Posts: 373
01 Oct 2014 01:57 AM

Looking into some of the data we're getting from Rosetta and PrimeGrid for your account, I think a clue has appeared that may explain your case of "CPID ping pong." Your Rosetta and PrimeGrid accounts have the exact same "created" timestamp, so it's possible that the BOINC client is somehow indecisive about which one to go with as the authoritative CPID.

Are you using the Charity Engine client software or a version of BOINC? Would be good to know version number, too. It might be that this issue can be addressed there at some level.

Dennis Nawrot ID: 6360 Posts: 32
02 Oct 2014 04:47 AM

BOINC 7.2.33 and BOINC 7.0.27

I have detached the second machine, but the ping pong seems to go on.


Tristan Olive ID: 22 Posts: 373
02 Oct 2014 04:54 PM

Ok, I think we have a way to test whether this is a bug in the BOINC client. (I'll be back with more information when we have it.)

Dennis Nawrot ID: 6360 Posts: 32
02 Oct 2014 05:25 PM

Very well.

Just in case it does matter, BOINC 7.0.27 is the still active machine.

Tristan Olive ID: 22 Posts: 373
02 Oct 2014 05:26 PM

The "created" timestamp on your PrimeGrid account has been tweaked to be one second later than the Rosetta account timestamp. Now I think we just have to wait and see if: 1) this fixes it; and 2) it stays fixed!

If so, we can look into fixing the apparent bug in the client.


Tristan Olive ID: 22 Posts: 373
02 Oct 2014 05:26 PM

(and thanks for the version info in that regard)