I completed a number of results for SETIboinc, but they all seem to have "Client Error" status, even the ones that have not been uploaded to the server yet.


Results 40694294, 40702556 and 40702557 are not yet returned(!) and the client gets "Failed to upload" mesage and then "temporary backs off".



I also crunct for Predictor end Einstein, and there, every result I returned is valid. Been crunching SETIclassic for 2 years now, but if the BOINC version of the project is going to reject each of my results, I may as well give up on SETI...



Results here

EDIT: Just realised that I could simply give a link to my result page instead of copy-n-paste.

Stanislav,

Every single wu is taking 0.00 seconds...unusual.

And inside of every WU is information about what your computer thought about it.

This is some info about your result #40702557. (I just clicked on it in your list of WUs)

40702557
[core_client_version]4.19[/core_client_version]
[message]process exited with code 13 (0xd)
[/message]
[active_task_state]1[/active_task_state]
[signal]0[/signal]
[stderr_txt]

[/stderr_txt]

So if you look for "code 13" among these message boards I think you will find a discussion about why, perhaps.

The thing is - none of the results took 0 seconds! - each result took anywhere between 3 to 4 hours to complete. I'll chect code 13 out...

EDIT 1:

... found threads on code 2, code 137, code 127, but not code 13 :(

I also saw that some of the results had "Couldn't start the app for this result: error -108" I saw these messages in my logs, but the result was always restarted a short while after that and completed to 100% (a project is removed from memory, when another project is scheduled to run). As I mentioned earlier, it is only SETI that I have these issues with.

EDIT 2: Managed to actually upload 2 of the 3 results with the "preempted" errors, not that it changed their status...

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

>
> ... found threads on code 2, code 137, code 127, but not code 13 :(
>

Code 13 is the dreaded segmentation / general protection fault.

It means that your software tried to access memory not belonging to it.

Two possible reasons:

Buggy hardware - Did you overclock your PC?

Buggy software - since seti runs fine otherwise, you should try to detach
and reattach to seti, to get a new copy of the seti app.

Regards Hans

---

Seg.fault?! Argh!!! Hardware is NOT overclocked. I have a P4 3.2GHz HT machine, which runs at factory hardware settings (FujitsuSiemens). I'll try resetting the project, though it is a pain on a modem line...

>
> >
> > ... found threads on code 2, code 137, code 127, but not code 13 :(
> >
>
> Code 13 is the dreaded segmentation / general protection fault.
>
> It means that your software tried to access memory not belonging to it.
>
> Two possible reasons:
>
> Buggy hardware - Did you overclock your PC?
>
> Buggy software - since seti runs fine otherwise, you should try to detach
> and reattach to seti, to get a new copy of the seti app.
>
> Regards Hans
>
>

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

> Seg.fault?! Argh!!! Hardware is NOT overclocked. I have a P4 3.2GHz HT
> machine, which runs at factory hardware settings (FujitsuSiemens). I'll try
> resetting the project, though it is a pain on a modem line...

Did you look in your machine to see what the error code might be, or did you get the error 13 from the forums?

---

> Did you look in your machine to see what the error code might be, or did you
> get the error 13 from the forums?
>

No, I read the error code in the results page (e.g here). I've reset the project before the weekend downtime and am now downloading the new 4.24 boinc frontend and the 4.09 S@H backend.Hopefully, things will be more reliable this time. I'll post again if there are still problems...

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

You can run Memtest86+ to see if you have problem with memory modules.

---

SETI@home classic workunits: 1,985 CPU time: 24,567 hours



Professor Desty Nova
Researching Karma the Hard Way

> You can run Memtest86+ to see if you
> have problem with memory modules.

Or Prime95 to test the CPU & memory intensively.

---

Grant
Darwin NT

Hey Stanislav! I just wanted to welcome you to BOINC, as I was a collegue team member in the classic S@H. I migrated to BOINC some time ago (last year), but it was a bit lonely in the UiO team then so I joined "Norway" team instead. :)

I hope you get these problems solved. Happy crunching!

---

Hei, Bjørn, thanks! :)

I see your point about the UiO team - if I manage to deliver an accepted result through BOINC, I'll be the only one on that group in this realm. Probably more will come as the classic verion shuts down... Oh, well, team Norway looks interesting, so I'll just switch to it in the mean time... :)

To Grant and Professor Desty Nova: I'll try those out if the results continue to be botched after the reset. Predictor and Einstein projects run flawlessly, though, so I somehow do not think the problem lies with the memory modules.

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

Returned 2 results using core client 4.25 with status success and already got credit for one of those. As I am pretty much disappointed with the front-end of 4.25, I again downgraded to 4.19. Now it remains to see if it was the core client or the back-end that caused the problem.

Just a thought - when the backend (setiathome 4.09 in this case) is downloaded is it compressed or protected by any checksum mechanism? Could it be that it was corrupted while being downloaded the first time?

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

Glad that you finally got it to work! I haven't upgraded to v4.25 yet either. The reports in these forums are mainly negative about the 4.25 release, so I stayed with 4.19 for the time being. You may want to check out BoincView: http://boincview.amanheis.de/?page=download

About the checksum, yes, the file has a MD5 checksum. (See download page: http://boinc.berkeley.edu/download.php) and this could very well be what you experienced. Since you used a modem to download while the server load was so high, it may have corrupted the file. Just a theory...

So, have you joined team Norway then? ;)
I see you have got yourself a stat box and an avatar, so you should be set! With the machine park you have, you'll be rushing past me in no time... LOL

---

> Glad that you finally got it to work! I haven't upgraded to v4.25 yet either.
> The reports in these forums are mainly negative about the 4.25 release, so I
> stayed with 4.19 for the time being. You may want to check out BoincView:
> http://boincview.amanheis.de/?page=download

Trying out it now... looks interesting. Might as well be upgrading back to 4.25. The main issue for me in 4.25 was the GUI, so if I can completely circumvent it my using BOINCView, there is no reason not to upgrade...

>
> About the checksum, yes, the file has a MD5 checksum. (See download page:
> http://boinc.berkeley.edu/download.php) and this could very well be what you
> experienced. Since you used a modem to download while the server load was so
> high, it may have corrupted the file. Just a theory...

Yeh, there is the checksum for the client, but does the client itself control checksums, when it downloads WUs and back-end updates?


> So, have you joined team Norway then? ;)
> I see you have got yourself a stat box and an avatar, so you should be set!
> With the machine park you have, you'll be rushing past me in no time... LOL

Yepp, I am on the team and settled down :) Though I don't think I'll be rushing past you any time soon - I had a park while I worked on a project, but that was a year ago. Now, it's only my PC doing the job... With the other projects, I think I'll be able to do about 60 credits a day.

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

> Yeh, there is the checksum for the client, but does the client itself control checksums, when it downloads WUs and back-end updates?

I don't know... Maybe you'll have to check it manually. You may find the answer in Paul's excellent BOINC documentation:

http://boinc-doc.net/index.php

> Though I don't think I'll be rushing past you any time soon - I had a park while I worked on a project, but that was a year ago. Now, it's only my PC doing the job... With the other projects, I think I'll be able to do about 60 credits a day.

Ok, that's a relief! LOL
No, just kidding. Welcome to team Norway! :)

---

I was of two minds wether to post it on this thread or on the once concirning ver. 4.25. Decided on this one as there are fewer messages to load before reply box and it consititutes a status update :)

After I went beck to 4.19, I produced another dead result, so I decided to "bite the sour apple" (a verbouse translation f a Norwegian saying :) ) and reinstall 4.25 botched GUI or not. Funnilly, on 4.25 thread, there is a post stating that the user returned garbled results with 4.25, but for me the opposite seems to be the case - probably the raw data in the workunits had errors?

Anyway, this time I installed 4.25 as a service and chose not to have the manager starting on login. Coupled with BOINCView 1.0 this seems to be the best solution - to have the daemon with all the bugfixes and to avoid seeing the awful GUI of the manager. BOINCView should really be made the default manager on Windows platform - elegant, simple, yet powerful. (One feature that I especially liked is to refuse downloading new workunits for selected projects - useful if you use the PC more than on average and is unable to meet the deadlines with what you have, never mind the new work)

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>

> I completed a number of results for SETIboinc, but they all seem to have
> "Client Error" status, even the ones that have not been uploaded to the server
> yet.

I've seen this behavior as well. Since I installed cc 4.25 those kinds of errors started to show up (4 or 5 WUs) - could be a coincidence though. All of them toke around 5 hours to complete and credit is claimed. When I return a result and the outcome is equal to “Client Error" why is credit claimed? I seems strange to me!

Be happy!

---

Happy crunching

Glad you found a working solution, Stanislav. Thanks for the install info BTW, it's very useful for us to know before upgrading.

I haven't upgraded to 4.25 yet, and I'm unsure if I will or wait until the next version (unless they pull the plug on 4.19 and force me to upgrade). Maybe the next stable version has ironed out some of the critizized points (in 4.25) from users. Besides, I'm a little concerned about several reports of serious trouble with this release. So, the "stable" version will have to be beta-tested some more without my participation, I think. :)

---

[snip]

> Besides, I'm a little concerned about several reports of
> serious trouble with this release. So, the "stable" version will have to be
> beta-tested some more without my participation, I think. :)

Trouble? What trouble? :) Seeing as I run the beta version of Opera broweser, and been doing it for a while (since ver 7 came out), I can say that I am the type to take unncessary risks. :)

While I am at it - highly recommend Opera to anyone out there (www.opera.com).

---

---
<img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=875&amp;trans=off" />
Everything is just <A HREF="http://heim.ifi.uio.no/~stanisls/fysisk/">A Question of Physics</A>