F@H Hangs after a little while

[LOUISSSSS]

so i've been running my dual SMP's of F@H + FAHMON on my Q6600 for the past few weeks and its been going smooth. recently i saw an error on FAHMON saying "*HUNG*"

i restarted it and it kept going.

now, i'd start up both instances of F@H and it'd work (i know its working bc i see the % go up and i see the CPU usage at 100%) for a few minutes then stop advancing. I then see my cpu usage drop down to 0% @ idle...

any reason why this is happening?

i've restarted my 2 instances of F@H and my PC numerous times to try and fix this

 

[Insidious]

Here's a thread At the Folding @ Home forums that talks about some possible causes of a hanging client

Hope it helps

-Sid

:beer:

 

[Insidious]

I usually just re-boot (to make sure all F@H things are stopped) and then re-start the clients when Windows comes back up. Some times it works, sometimes I loose the workunit on an error.... I figure if Stanford cares about loosing these WUs they'll make more of an effort to solve the issue that causes this.

:beer:

 

[Gravity]

I agree sid. I read the thread and it sounds like crap. A little more investment in time. I also hate the barb about the bug being fixed in Vista.

Gravity

 

[LOUISSSSS]

wait, so u guys get Hangs every now and then too?

i think i can estimate i "hung" Client every 24 hours...

this doesn't have to do with system instability does it?

 

[Insidious]

Originally posted by: LOUISSSSS
wait, so u guys get Hangs every now and then too?

i think i can estimate i "hung" Client every 24 hours...

this doesn't have to do with system instability does it?

Well, you really have to be watching your systems to answer that question.

If I can't find instability any other way (ie: gaming, Prime95, memtest) then I assume it is probably network or just Stanford shortcommings.
Usually when I have pushed an overclock too far, my symptom has been a spontaneous re-boot after a few hours or even a couple days. If I see any of that, I back things off a little.

I don't get nearly so many hangs as you are describing, but I keep most of my PCs on a wired network connection. It is the wireless (especially quads with two clients running) that is most prone.

-Sid

 

[LOUISSSSS]

yea an unstable OC has crossed my mind, but i dont think it should be all of a sudden. my 2 x SMP clients have been running fine with the affinity changer for about 3weeks to a month before it started hanging everyday at least once or twice... this really slows down my work since it usually happens when i'm afk for a long time (at school or asleep.)

and F@H only uses the network when downloading an entire WU and after finishing an entire WU right? nothing inbetween so it shouldn't be my network disconnecting (if i even do get D/C @ like 40% of a WU for example)

Stanford shortcommings...? what would u mean by that

 

[Gravity]

I have developed a strong disdain for the smp client. It needs much care and is typical of "beta" software. I wish they would work on it a little more.

Frustrated but folding,

Gravity

 

[ch33zw1z]

I pushed my E6600 to about 3Ghz and encountered something similar. Running @ stock now, and happens much much less. Odd, thought it was just me

 

[Gravity]

I'm overclocked a bit too but several of my other boxes that are overclocked are folding like banshees; assuming a banshee has some l33t folding skillz.

Gravity

 

[Markfw]

Well, I have one of the currebt 7 that hangs about once a month at the most. I don't know if I am lucky or what.....

 

[Gravity]

Somehow I doubt that's luck. I want all these boxes working hard this weekend. I should be close to 20k ppd if they all do their work.

Bastids they are,

Gravity

 

[Insidious]

As far as I can tell, a wireless network is the biggest contributor (by far), next is two instances of SMP, and last is overclocking. (I also believe some installs of WinblowsXP are inherantly less stable than others (depending on the myriad of things we love to install, uninstall, etc.)

To Louis, the networking on your computer is used constantly (loopback adapter) when running a SMP client. not for communication to the outside world but for communications between the threads running F@H (Good idea Stanford..... NOT).

I don't see nearly as many hangs as I have heard others talk about, but it is always a little bit of a crap shoot when I've been installing stuff and have to re-boot..... maybe F@H will work right next time.... maybe not. (it's usually OK).

I do know that on my PCs with two SMPs running, if I use Cntrl^C to kill one instance it is not unusual at all for the other instance to bork itself. (another WTG for Stanford programmers) For this reason, anytime I need to kill F@H on a 2-SMP PC, I just do a normal re-start of Windows.... it is much more relliable for me.

-Sid

BUT...... I DO run SMP and I like Folding @ Home, so don't read too much negativity in my posts.

:roll: