Hello!

BOINC 4.19 just downloaded the 5th CPDN model after the third aborted with the traditional Error Code -5 way before its time.

I seriously asking myself if it makes any sense to run CPDN on this machine (P4 2.8 GHz HT, WinXP SP2).
I do turn on and off the machine quite a bit (2x a day) and maybe that is something CPDN can not handle? At least that is the only reason I can see why it should keep failing.
BOINC is the only software on my computer having trouble. It is an extremely stable system, so I tend to blame CPDN for the problems.

I would appreciate any feedback from the community on whether it makes sense to run CPDN under these circumstances or which other project can handle the frequent on/off switching.
And: Yes, I do supend and exit BOINC before turning the machine off!

Thanks!

Friedrich

---

I love CPDN!
--

Hi Friedrich

Make sure that when you suspend and exit, you give it time - it does not respond very quickly to these instructions.

Other advice that has regularly been given to people who have this same problem is to check your machine for stability, by running stability tests and checking for overheating, even if it runs stably with everything else. This is because cpdn is a very demanding programme.

My boinc cpdn models constantly crashed with the same -5 error code, which is a calculation error, even after I uninstalled, reinstalled, upgraded etc. Eventually I concluded that my system is incompatible with boinc cpdn at the moment and adopted the solution of last resort. This is to uninstall boinc cpdn and install classic cpdn instead. This is running perfectly at about the same speed as a boinc model, trickles regularly, and the last finished model uploaded automatically. At the moment classic models are equally useful to the researchers.

This is a method allowing continued participation in a most valuable project. After completing another model or two I may try boinc cpdn again.

But look first at the many posts (on both boinc message boards) about -5 crashes!

---

__________________________________________________

Hello mo.v

> Make sure that when you suspend and exit, you give it time - it does not
> respond very quickly to these instructions.
I do that. I usually wait until the icon disapears and then check the task manager. Does not make a difference.

> Other advice that has regularly been given to people who have this same
> problem is to check your machine for stability, by running stability tests and
> checking for overheating, even if it runs stably with everything else. This is
> because cpdn is a very demanding programme.
I don't think it overheats since my Intel monitor software tells me that I am well within safe margins. Also I then should see other effects with other software while running CPDN. But everything else is running fine.
However, all references to what kind of stability tests I could do point to the phpBB which is currently off-line. But all that would tell me is that my system is unstable (if it really is), which does not change anything.

> moment and adopted the solution of last resort. This is to uninstall boinc
> cpdn and install classic cpdn instead. This is running perfectly at about the
> same speed as a boinc model, trickles regularly, and the last finished model
> uploaded automatically. At the moment classic models are equally useful to the
> researchers.

That might be the right solution for me, too.

> This is a method allowing continued participation in a most valuable project.
> After completing another model or two I may try boinc cpdn again.

I am going to wait for a more fault-tolerant version of BOINC/CPDN.

> But look first at the many posts (on both boinc message boards) about -5
> crashes!

I have read a lot of them, but I could find nothing in there that really helps.
So I will wait for the next crash and then switch.

Unless somebody else has a good idea...

Thanks!

Friedrich

---

I love CPDN!
--

I get this quite frequently, but only after I shut down BOINC, restart BOINC, and then wait for CPDN to start. It doesn't sem to matter how long I wait for CPDN to completely shut down. I suspect that there is a problem in the checkpoint code.

---

BOINC WIKI

> I get this quite frequently, but only after I shut down BOINC, restart BOINC,
> and then wait for CPDN to start. It doesn't sem to matter how long I wait for
> CPDN to completely shut down. I suspect that there is a problem in the
> checkpoint code.

Checkpoints are saved at every 144 timesteps. Just to follow on to what mo.v said to ensure that in the boinc front end one suspends, then exit boinc. It does sound as if you are both trying to shut down properly but something is happening to the files being written.

If you are having problems running CP one thing to check is the stability of your machine with software such as: <A HREF="http://www.memtest86.com/">Memtest86</A> and <A HREF="http://www.mersenne.org/freesoft.htm">Prime95</A> torture test. These software will test your system to help show if any problems exist. Your problems appear to be with the shut down phase though.

Another think to check is that the hard drive is defragmented. To do this you should exit boinc first :)

It could be a problem with some of the controling xml files or the restart information written by CP but I am out of my league there. If you have another PC that runs CP OK you could compare some of the files to see if you can find a problem but I don't know if that would help.

&gt; I get this quite frequently, but only after I shut down BOINC, restart BOINC,
&gt; and then wait for CPDN to start. It doesn't sem to matter how long I wait for
&gt; CPDN to completely shut down. I suspect that there is a problem in the
&gt; checkpoint code.


If you kill the remaining CPDN processes immediately after shutting down BOINC, the CPDN models might survive.

I tried both, letting them continue and killing them. Letting them continue always resulted in a damage report to the server, killing them always left them intact.


So it seems to be a bug in the process handling, not in the checkpoint code.

Unfortunately I do not always remember to check for running tasks everytime.


But never kill the CPDN tasks _before_ you stop BOINC as BOINC will rat on you at the CPDN server and the model will be gone too.