Thanks for the info, Bobcat, and for providing the links. Later today I\'ll see if there\'s any way these people can be contacted. I\'ve mentioned this business about current models needing boinc version 5 in the cpdn news on all 3 cpdn forums plus two stats forums, but if these members aren\'t looking at their boinc manager or models graphics (and finding they\'re not there), they\'re unlikely to be watching our news threads.

---

Cpdn news

Admin Milo in Oxford has now composed a template for an email that can be sent to many if not all these people. We can also now send a private message to those who aren\'t called User or Administrator. Saves a lot of time.

---

Cpdn news

I sent 10 or so PMs to the Mac Darwins getting error code 6 (the shared memory issue). Hopefully some of those PCs might get sorted out now. A few more Macs were \'anonymous\' (no way of sending a PM).

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

A bit late to add a comment but I ran into this problem myself back in June. I don\'t know what caused it but while I was watching it I saw CPDN download 2 WU\'s on a single core. I immediately went and told it to not get any more work and that\'s how I keep all my systems until it\'s ready to get more work. I haven\'t seen the problem reappear even after I\'ve enabled work (gets) since then.

PS... I aborted the second WU so that it didn\'t sit on my system for any length of time before it would be sent out again. That system just finished the WU I got in June about a week ago. I registered in Oct last year but didn\'t start processing full time on all 3 computers until June. Since then I\'ve completed I think 3 or 4 models and had one crash with less than 200 hours left. <sigh>

---

Hi Arion

I don\'t think that tasks/models or projects are allocated a specific core. So if you only had one climate model and allowed network activity and in your project preferences had said that cpdn could use both cores, then the opportunistic cpdn server would try to get two models running on your computer.
You did all the right things.

Ways to prevent models crashing are set out in the project READMEs - the best item for this is #5 by Mike in the README about avoiding crashes and problems. It\'s been recently updated and it\'s also a sticky near the top of this Number Crunching section. Or get to all the READMEs thro Mike\'s sig or mine. If you possibly can, you need to back up the entire boinc folder regularly; this is the only way to get a crashed model back and running for you again. With such long models none of us can guarantee that they\'ll never crash.

The problem with most of the members mentioned in this thread who are downloading multiple models is that they still have v4 of boinc and the models won\'t run on it at all. A lot of these people have received emails from Milo and I have another list of host IDs to give to him. And at some point I\'ll also check to see who needs a second email.....

All these people need to do is upgrade boinc, or set CPDN to No new tasks, or detach from CPDN. These easy solutions are all suggested in the email. Most of them have got into this mess because they never look in on the forums or post.

Which of course is not your situation at all.

If members find records of computers still downloading multiple models, it\'s by no means too late to post a link to them in this thread.

---

Cpdn news

I apologize for bringing such an old thread back to life, but I have found another host that is erroring all tasks and has accumulated 0 credit.

Host 884765 has over 660 tasks in error. It appears to be a 64-bit linux system, so it may just need the 32-bit libraries installed to make it work.

. . . has over 660 tasks in error. It appears to be a 64-bit linux system, so it may just need the 32-bit libraries installed to make it work.

That\'s probably the answer. This has been a recurring issue.

Edit: Should have mentioned, I sent a PM to the participant. (I hope he/she allows messages from the Boards.)

Thanks for calling this to our attention.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

An ongoing problem which we\'ve mentioned several times in the forum News threads is that the BOINC default for project accounts is to have email notification of private messages turned off. This is to protect members\' privacy and I can see why BOINC makes no email notification the default. But it\'s a thorough nuisance for moderators and other members who try to inform people that, for example, all their models are crashing. (Some members don\'t realise that their models are crashing out.) Unless these people visit the forum they never see that they have a private message waiting.

I\'ve heard that only 3% of BOINC crunchers ever visit the forums. I don\'t know how that was ascertained, but the figure doesn\'t surprise me. This situation means that most members of most projects are effectively incommunicado.

Milo and Tolu, however, have access to members\' email addresses (the moderators have no access) and now have a script that detects computers that crash a particular number of models in a week or month. These people are automatically sent an email and invited to ask for help on the forums. I can\'t remember offhand whether these computers\' supply of new models is then severely limited or cut off altogether.

But if members notice other people\'s computers that are crashing lots of models and may have slipped past the script detection, they are more than welcome to report this on the forum. The mods can then ask Milo to send the email manually.

---

Cpdn news