Slowly I'm going nuts with CPID line-up. All of my 5 BOINC projects have 2 different CPIDs, let's call it CPID1 and CPID2.

E@H, CPDN, P@H and LHC all share CPID1, only S@H uses CPID2. After contacting S@H at work everything seems fine, CPIDs are lined up, all Project share CPID1.

BUT: Connecting S@H from my home PC S@H CPID ist switched back to CPID2.

EMail adresses, user names etc. are identical, even the OS/Client is the same (Win2k, core client 4.25). Reattaching the project didn't work.

Any way to fix this CPID wierdness? Otherwise I have to detach from S@H on my home PC.


scsimodo

---

hello scsimodo,
what else are you running @home?
And what CPID has it after the next connection?

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

> hello scsimodo,

Hi Sänger!

> what else are you running @home?

Attached to all Projects but ATM only Predictor is active (and Seti, if I try to connect). Same @work + active Einstein

> And what CPID has it after the next connection?

The same (wrong) one. No change @home, @work CPID lines up with the very first connection. Strange ...

Just saw another oddity. CPIDs showing on the project websites are different than those in client_state.xml. Shouldn't they be the same?

Hmm, I'd better deactivate Seti until I found a solution. Now that I finally have combined stat at boincstats.com again :-)

scsimodo

---

> > hello scsimodo,
>
> Hi Sänger!
>
> > what else are you running @home?
>
> Attached to all Projects but ATM only Predictor is active (and Seti, if I try
> to connect). Same @work + active Einstein
>
> > And what CPID has it after the next connection?
>
> The same (wrong) one. No change @home, @work CPID lines up with the very first
> connection. Strange ...
>
> Just saw another oddity. CPIDs showing on the project websites are different
> than those in client_state.xml. Shouldn't they be the same?
>
> Hmm, I'd better deactivate Seti until I found a solution. Now that I finally
> have combined stat at boincstats.com again :-)
>
> scsimodo
>
4.19 and 4.25 have different algorythms for determinng which of the two CPIDs to keep. Make certain that all your machines are on the same version. 4.19 would take the numerically largest, 4.25 takes th CPID from the project with the oldest join date.

---

BOINC WIKI

> 4.19 and 4.25 have different algorythms for determinng which of the two CPIDs
> to keep. Make certain that all your machines are on the same version. 4.19
> would take the numerically largest, 4.25 takes th CPID from the project with
> the oldest join date.

Same OS (W2K), same client (4.25).
Even a complete reinstall didn't fix it.

---

I noticed another question. No, the CPID in the XML files on your computers is the Internal CPID. The CPID that all of the stats sites get and that some of the projects display is the external CPID. To get the external CPID, hash the internal CPID with the email address (without doing anything to the email address such as changing the case). So there are two versions of the CPID.

To fix the problem, I would detach from S@H on the problem machine, make certain that the CPID is correctly updated from one of the other machines, and then re-attach S@H from the problem machine. The other way is to copy the CPID around yourself in the xml files on your problem machine - just remember to turn BOINC off first.

---

BOINC WIKI

I suspect there may be really be something strange going on with the CPID. The code has recently been updated to work in a way that is almost opposite to how it worked before. I have also had problems getting it to syncronise, and found a solution.

All my hosts are running BOINC 4.25. I have never run a prior version. At present, one host is linux, the rest are windows. All run S@H in combination with at least one other project. One windows machine runs ALL projects and was the first machine I set up (holodeck).

I originally signed up with Einstien before I activated my S@H account since S@H was offline at the time. Then I signed up to all other open projects on the same machine before I deployed boinc on the other hosts, so theoretically the first machine should have syncronised the CPID across all projects, and it should have propogated to the additional machines. All projects use the same email addresss, which is in lowercase.

Ive checked out boincstats.com over the past several days, and for more than 5 days, with all machines constantly in contact with the schedulers (at least once per day for each project) the CPIDs did not syncronise automatically. What I found was the same as scsimodo (the original poster). ie: There were 2 CPIDs. S@H had one CPID, all the rest had the other CPID.

The 'My Account' page for S@H showed the same CPID as on boincstats, as did the account page for the other projects. S@H was the odd-one-out.

Finally it bugged me enough to go crawling around inside client_state.xml and what I found was that the CPID and email hash entries in the project section of the other projects were all the same. S@H had the same email hash, but the CPID was null. In other words, the S@H CPID line read:

<cross_project_id></cross_project_id>

I stopped boinc, copied the CPID line from another project section to the S@H section, restarted boinc and updated (twice) and then checked the S@H account page. The S@H CPID was now the same as all the other projects. Yay! (note: Im glossing over the difference between internal and external CPID here).

I updated the S@H project on the other hosts and checked the client_state.xml files, and they all now had the proper CPID for S@H. (They too were previosuly null - the only place that actually held the weird CPID was the S@H website).

I fully expect boincstats to show a single CPID in the next couple of days.

I hope this helps anyone who has been pulling their hair out over this, and I hope that it also helps the devs find this bug, and kill it.

[Wishlist Item: - Message boards should automatically convert < & > to &lt; & &gt; Each time I edited this post, I had to retype the example above because it 'dissapeared' into the html code.]

Same here, two CPIDs that won't line up.

Most (but not all) hosts have the same {cross_project_id} entry in the reply and state XMLs for all projects - it's the one that CPDN propagates (no hashing with email needed). A few hosts/projects have totally different ones - all on 4.19.

There is a path between all projects but still it doesn't change the CPID on Einstein, Seti and Pirates.


Could it be that CPDN uses a different server side algorithm for the CPID ?

---