Finally Folding [H] Again

[CaptRingold]

Little things sometimes brighten my day, and knowing my 11 Ghz worth of CPU's I have under my general control ( 6.5ghz of my own, and my girls 4.5ghz) are all, or about to again, be folding at least part-time for the [H]. I, of course, make my machines gruel on all through the lonely night, but the girlfriend, neh, she doesn't make her computers work for their keep in this world.

Anyway, I DO have this computer overclocked, an A64 3200+ @ 2.45ghz, but.. well, it appears stable, no Memtest problems after 6 hours, no Prime95 errors after 48 of off-and-on (almost all on, 1 30hr-ish solid stretch), and most everything seems to work well, except Futuremark's stuff, but I chalk that up to being in Win x64. I figured my final stability test would be to closely watch the FAHlog.txt for issues. It went through a Gromacs core just fine, but after it got to the end:

[03:09:03] + Attempting to send results
[03:10:16] - Server reports digital signature does not match.
[03:10:16] (May be due to corruption during network transmission or a corrupted file.)
[03:10:16] Could not transmit unit 01 to Collection server; keeping in queue.
[03:10:16] + Closed connections

No errors while doing the work, mind you, just this, repeatedly, as it tried several times to send the workunit. In fact, I think it's still trying to send it, even though it's working on another wu at the moment.

Ideas? If I thought it was stability, I'd run another battery of Prime95 and Memtest tests, but I've had stability problems before, and usually saw a WU crap itself long before it ever finished, so not sure here.

 

[KodiakStar]

Start with trying to use the -sendall flag like they talked about here:

http://hardforum.com/showthread.php?t=909025

That _should_ work, if not get back to us..


And welcome back

 

[DR_K13]

Welcome Back

 

[SmokeRngs]

I highly doubt that it has anything to do with an unstable machine. That's normally not something you would get from one. As the error indicates, it looks like it was a problem in transmission. It could also just be a bug or something with Stanford's servers.

One 210 point Gromac I had plodding along on my old P3 700 system at work finally finished and the machine tried to send it back. Stanford refused to take it because it said it had no clue where the hell the WU had come from. It sat in queue and the client tried to send it several times a day until the day after it was due. Then Stanford accepted it back but gave me no credit for it. I was not exactly happy about this but there wasn't much I could do. Also, I had tried the flag "-send all" and that didn't help. It was one of the first things I did and it gave me the same error as any other time.

I would suggest leaving the WU alone for now. The client should try to send it back every few hours plus whatever other connections needed to send back or download new WUs. You'll probably know in a few days whether it's going to take it or not.

Hopefully, someone will have a better idea of what needs to be done other than wait.

 

[CaptRingold]

Thanks for the welcome back guys, I'd been fairly annoyed most my systems haven't been folding for a while now, but at least they're back to it. I'll go click on this worthwhile link to make myself feel a little better about the lost points

I'll try that flag, and see what happens. If it ends up being like what you said SmokeRings, then I'll let her sit for a couple days and search Stanfords forums to see if they've had the bug reported before to make sure they're aware of it.

 

[CaptRingold]

Well, client has finished another work unit, a smaller and quicker one, and had a hard time sending it too. So, I closed the client (using the graphical one on this particular comp), added the "-send all" flag, and launched it again. Gave me this:

[01:17:47] + Attempting to send results
[01:18:04] - Server reports digital signature does not match.
[01:18:04] (May be due to corruption during network transmission or a corrupted file.)
[01:18:04] Could not transmit unit 02 to Collection server; keeping in queue.
[01:18:04] - Failed to send all units to server

Folding@Home Client Shutdown.

Client shutdown? I guess after all that work being rejected by a big mean server, it got depressed.

EDIT:
Turned off Nvidia's Active Armor firewall deal thing that my board has, and tried it again. And, it worked. When the first unit finished, my current settings had been 'Medium', and todays unit finished with the firewall on 'High'.. I think it's just blocking something by default, but not sure what: any ideas on that? I'll create a rule allowing the winFAH.exe to access the internet, see if that has an effect next time.

 

[Mattman]

EDIT:
Turned off Nvidia's Active Armor firewall deal thing that my board has, and tried it again. And, it worked. When the first unit finished, my current settings had been 'Medium', and todays unit finished with the firewall on 'High'.. I think it's just blocking something by default, but not sure what: any ideas on that? I'll create a rule allowing the winFAH.exe to access the internet, see if that has an effect next time.

That thing is trouble. I had to reformat my machine after it decided that it didn't want me communicating with other machines on my network and wouldn't let me install/reinstall/uninstall it. After researching my Nvidia firewall problem, I now recommend to others that they don't install the Nvidia IDE drivers or the firewall (Nvidia access manager) on their Nforce boards. Other than to say "avoid it at all costs" (until Nvidia releases working drivers, that is), I can't help you out much. If you're lucky, Nvidia access manager will allow you to uninstall itself - if not, it's time for a fresh Windows installation.