LHC@home

Sorry guys, this seems to be a problem with the core client; I don't think there is anything we can do about here at LHC@home.
Thanks for participating.

/Jakob Pedersen
LHC@Home Admin.

As i asked in Memnoch's thread in questions and problems, has anybody told the Boinc devs about this email or message post, as they are the people who really need to know about this to able to fix the problem.

I had LHC taking over one of my computers. I did the following things and it cleared up.

1) restarted computer
2) looked at client_state.xml and the acount*.xml files. They seemed to have the right information except the name of the projects was LHC.
3) restarted client
4) restarted computer again
Still showing LHC on most projects at this point.
5) updated all the projects showing LHC.

It seems to be working fine now (about 2 hours later).



John Keck -- BOINCing since 2002/12/08

I have had machines overrun with pirates and LHC ...

The site I seem to have most of my download errors from is LHC@Home. Why it seems to want to redown load the science appliction is beyond me ...

Ohhh ....

could that be because ... if the BOINC Work Manager is messing with the project names it does not get a match on the current science application and tries to get a correct one, adds/changes the files with the new science application name but when it does the compare, the project NAME is wrong so it tries again?

Usually, getting a scheduler hit off of update seems to work ...

Reboot alone does nothing ...

Can you folks at the server side see the hits on downloading the application?

Off topic,

with these real short WU, I am hitting my daily quota. You may want to consider upping that or making them longer ... Just a thought guys ...

Or, mix and match with long ones and short ones ...

<p>
For BOINC Documentaion: Click Me!

@Paul Buck
you write
Or, mix and match with long ones and short ones ...
-------
i have this crunching times are 6 minutes 15 min. 30 min and some with 1 hour
on a amd64
------------

feel free to visit Guidos Boinc Forum
Better to know now if there is a problem than to have an angry mob of users beating at your door.Investigations continue.;-)))

sheesh... This message got nurffed by the system! See above.

Early this AM (0500 Pacific) I had another machine go nuts.

I applied the fix that I detailed below and all seems well now.

The second PC freaked out as soon as it downloaded siztrack 4.45 (it had 4.43 before).

Sorry, Girlieman, but this does not look like a BOINC core app issue. If it were BOINC, how can it be explained that this happens only with the LHC sixtrack 4.45 app installed?

Here is my current "scorecard:

4 PCs running BOINC 4.09 on W2K Pro SP4 + all updates

2 PCs running sixtrack 4.43 and are OK

2 PCs running sixtrack 4.45 and have freaked

It should be noted that after doing the "fix" I detailed in an earlier message that both of the 4.45 machines seem to be OK. Also note that this "fix" has only been tested for ~19-20 hours so far and may not be the end-all solution.

Based on the fact that my "fix" and that other users have replaced the client_state.xml file and both seem to get things working again, that makes me think that there is not so much of a problem with the sixtrack 4.45 app as there may be an issue with the "upgrade" from 4.43 to 4.45, perhaps in the installer.

I am now waiting for the other 2 PCs to "upgrade" to siztrack 4.45 and go belly up. Yes, I have considered detaching from LHC before that happens, but to what end? How can the problem(s) be found if no one can provide the LHC team with information?

Anyway...

As far as the politics, I like being right... Wing. :)

Still strange...

The 2 "fixed" PCs are working OK and have shown no further problems.

Interestingly, one of the PCs that has (so far) shown no problems on sixtrack 4.43 just downloaded sixtrack 4.46.

Looks like LHC has figured out that there is a problem with the 4.45 that caused so many problems.

> [url=http:www.GeorgeWBush.com]
>

kc6ufm, could you please remove this link. The problem is not that much with the political content as with the fact that this picture seems to confuse my Mozilla browser such that the picture exceeds the table cell area.

For the client problems I can't say much at this point.

Markku Degerholm
LHC@home Admin

> I had LHC taking over one of my computers. I did the following things and it
> cleared up.
>
> 1) restarted computer
> 2) looked at client_state.xml and the acount*.xml files. They seemed to have
> the right information except the name of the projects was LHC.
> 3) restarted client
> 4) restarted computer again
> Still showing LHC on most projects at this point.
> 5) updated all the projects showing LHC.
>
> It seems to be working fine now (about 2 hours later).

When it first happened to me, I also looked at the client_state.xml and it looked like the project name was the only thing affected. I manually renamed S@H (which thought it was Pirates) and thought all was good. Then when S@H tried to return a result, I got an error saying it couldn't recognize the app version. So all might not be well, have you returned any results?

<a> [/url]

> kc6ufm, could you please remove this link. The problem is not that much with
> the political content as with the fact that this picture seems to confuse my
> Mozilla browser such that the picture exceeds the table cell area.

You got it...

I would be interested in what version of Mozilla you are using that causes problems. Mine doesn't (1.7.3) have an issue, nor does AOL (8.0 and 9.0), nor does IE (6.0). I even tried Netscape 4.73 and all was well.

No political agenda assumed... Now, as long as no other political banners appear in the groups, we are in good shape.


> For the client problems I can't say much at this point.

We'll have to wait and see if the 4.46 siztrack fixes/prevents the problem(s).


> Markku Degerholm
> LHC@home Admin

have firefox 0.9.3-all things work fine on project page
feel free to visit Guidos Boinc Forum
Better to know now if there is a problem than to have an angry mob of users beating at your door.Investigations continue.;-)))

> @Paul Buck
> you write
> Or, mix and match with long ones and short ones ...
> -------
> i have this crunching times are 6 minutes 15 min. 30 min and some with 1
> hour
> on a amd64
> ------------
>

They are issuing 100,000 turn models, I was suggesting that they also send out the longer ones too ... Just a thought because I am hitting the 50 limit, which seems to be by user instead of a more reasonable per computer. Some of the faster ones might blow through the work in moments ... :)
<p>
For BOINC Documentaion: Click Me!

I have told the BOINC devs about the problem, and they have identified the problem and are working on a fix. It seems that if two connections to different projects happen at the same time, BOINC gets confused and can copy the name of one of them into the other.

> I have told the BOINC devs about the problem, and they have identified the
> problem and are working on a fix. It seems that if two connections to
> different projects happen at the same time, BOINC gets confused and can copy
> the name of one of them into the other.
> <a> href="http://www.boinc.dk/index.php?page=user_statistics&project=cpdn&userid=13">
>

Rock on!

I reinstalled boinc from scratch again today.
No big loss.. wasnt' able to get work units from any project.


______________________________________________________________
Did your tech wear a static strap? No? Well, there ya go! :p

> I have told the BOINC devs about the problem, and they have identified the
> problem and are working on a fix. It seems that if two connections to
> different projects happen at the same time, BOINC gets confused and can copy
> the name of one of them into the other.

Thanks John. :)

You can also tell them it does more than copy the name. It starts downloading files into different projects folders. (after hijacking the name)

Heffed,

> Thanks John. :)
>
> You can also tell them it does more than copy the name. It starts downloading
> files into different projects folders. (after hijacking the name)

Given some of our past disagreements, it is good we have been agreeing almost all the time! :)

John,

I agree with Heffed! Thanks man...

I would have thought that they would have made a spin-lock on that ... but, I can see how easily it could have been missed. I guess with 5 projects, with only 3 of them up and sending work increases the risk/chance of seeing this problem.

I know in beta I said we need more cross-project testing... ah well ...
<p>
For BOINC Documentaion: Click Me!

> Heffed,
>
> > Thanks John. :)
> >
> > You can also tell them it does more than copy the name. It starts
> downloading
> > files into different projects folders. (after hijacking the name)
>
> Given some of our past disagreements, it is good we have been agreeing almost
> all the time! :)
>
> John,
>
> I agree with Heffed! Thanks man...
>
> I would have thought that they would have made a spin-lock on that ... but, I
> can see how easily it could have been missed. I guess with 5 projects, with
> only 3 of them up and sending work increases the risk/chance of seeing this
> problem.
>
> I know in beta I said we need more cross-project testing... ah well ...

They thought that did have the separate connections locked so that they could not happen at the same time....