WCG FAAH 4378 series issue

[nomad8u]

Since starting up with WCG, I've gotten in the habit of checking my results status on My Grid page about once a week filtering on Result Status "error" or "invalid" as a way to monitor my clients for problems. I found one yesterday from a run of FAAH on 9/25 that resulted in an error so I decided to investigate a bit.

It didn't take long on the forum to find the answer. I was pretty dang impressed with not only the speed of the response from the Community Advisors that reported it up the chain and the swiftness with which a "WCG Tech" responded with a resolution and explanation, but the overall professionalism of the resulting responses as well. Very refreshing. But I guess my biggest surprise came in the form of not even knowing I had a problem until I looked for it. The WU generated an error due to a problem with the WU itself and not a hardware issue, so it spent 0 processor time on it, uploaded the relevant info and got back to crunching the next WU.

I've been very impressed with the World Community Grid project as a whole and if we could just get our home stats page running (after the points conversion which looks to be their one wart leftover from consuming/converting from UD) I'd be one happy camper!

To anyone wondering, this is NOT a dig at F@H in any way as I'm still a participant there as well as WCG but rather a little face time for WCG as well as a positive look at how problems can be addressed and resolved.

To anyone else that has idle CPU cores for whatever reasons, I found what I believe are some worthwhile projects on World Community Grid. Help Conquer Cancer, FightAIDS @ Home and Discovering Dengue Drugs Together are just a few of the projects I participate in and think have potential to help our fellow man while giving some idle CPU cores something productive to do.

Keep Folding/Crunchin ON!

 

[relic]

I run f@h on my GPUs but since SMP was so buggy I stopped running any f@h CPU clients.

However, randomly the WCG client completely locks up some of my systems while others run with no problems. (The f@h SMP client constantly tanks on all of my boxen but never locks up the system) I like the WCG client, but I can only run it on systems that "like it".

So I am stuck with the following DC setup:

I run F@H on systems with GPUs that are capable of running the GPU2 client.
I run WCG on the quads that don't lockup with that client.
The quads that WCG locks up are idle.

 

[SmokeRngs]

I run f@h on my GPUs but since SMP was so buggy I stopped running any f@h CPU clients.

However, randomly the WCG client completely locks up some of my systems while others run with no problems. (The f@h SMP client constantly tanks on all of my boxen but never locks up the system) I like the WCG client, but I can only run it on systems that "like it".

So I am stuck with the following DC setup:

I run F@H on systems with GPUs that are capable of running the GPU2 client.
I run WCG on the quads that don't lockup with that client.
The quads that WCG locks up are idle.

Interesting, I haven't had a problem with the WCG client even once so far. I have it installed on a few single core systems including a P3 1Ghz, a couple of dual core systems, my X2@3Ghz and my sisters laptop with a Core Duo and I ran it on my main [email protected] for a while. I've also had it installed on a few other single core systems which have since been retired or pieced out. Not once have I had a problem with the system locking up or even the client crashing. This is a mixture of Windows XP and Vista machines as well as Ubuntu and openSUSE 10.2 and 10.3.

 

[relic]

Interesting, I haven't had a problem with the WCG client even once so far. I have it installed on a few single core systems including a P3 1Ghz, a couple of dual core systems, my X2@3Ghz and my sisters laptop with a Core Duo and I ran it on my main [email protected] for a while. I've also had it installed on a few other single core systems which have since been retired or pieced out. Not once have I had a problem with the system locking up or even the client crashing. This is a mixture of Windows XP and Vista machines as well as Ubuntu and openSUSE 10.2 and 10.3.

Just had to uninstall it on my Vista x64 due to the lockups. WCG runs fine on my Vista x86 Q6600 along with an F@H SMP client, so yes it does work fine on some systems.

It is definately the WCG client that causes the lockups (Tried 50%-100% usage) Shutting it down fixes the issue. This is the same system that was running f@h SMPx2 + GPU2.

Not sure what to tell you, I understand that you don't have a problem, but that doesn't help me.

 

[jws2346]

Like Mr relic posted above, I no longer run the WinSMP CPU client on any of my 24/7 dedicated folders. I've only gots three of them ATM (2x Q6600's and 1x E6600) I run one GPU2 client on each of my Q6600's (8800GT, 8800GS) and I run the LinSMP client on my E6600. (AFAIK there's no straight GPU2 client for Linux, when and if one does come along, not the Wine jobber, I'll say "sayonara Windose", my boxen are non critical and don't need the Windows OS ) It's so refreshing I can run the WCG clients on three cores without any problems at all (ie EUE's, etc) and on core #3 out of the four cores on a Q6600 I run the GPU2 client. I've run the WCG clients on three cores for quite some time trouble free (ie no EUE'ing, no video slow downs, no OC' problems, no "all of a sudden client expiration", etc and the CPU temps, read by Real Temp, are way lower )

I run LinSmp on a E6600 under Ubuntu 64 v8.04 or Kubuntu 64 v8.04 with a "dual booting" grub file. What decides whether I run LinSMP under Ubuntu or Kubuntu is pretty much decided on how I feel ATM. I think other than GUI's (KDE, Gnome) and a few other "out of the box" things, they're pretty much the same (ie synaptic, nautilus, etc)

When and if the GPU2 client becomes a problem (ie EUE's, point devaluation, goes the way of the GPU1 client or I have to invest large sums of money (ie more or different GPU's, PSU's, anything above and beyond the call of "common sense" ) to make it worthwhile to run my boxen 24/7 and burn the necessary electricity I'll just go 100% WCG

Foldin' and WCGn' for a CURE

 

[nomad8u]

Just had to uninstall it on my Vista x64 due to the lockups. WCG runs fine on my Vista x86 Q6600 along with an F@H SMP client, so yes it does work fine on some systems.

It is definately the WCG client that causes the lockups (Tried 50%-100% usage) Shutting it down fixes the issue. This is the same system that was running f@h SMPx2 + GPU2.

That's a shame. It seems like Vista x64 is pretty studly running some apps like the GPU client but not so much so with others.

Either way, it seems to be that Vista x64 and whatever Boinc client version you installed aren't playing nice. Curious if you know what the client version is that you were trying to run? If you downloaded it from WCG it's probably the WCG branded 5.10.45 version. If it was the WCG version, did you get the option for separate downloads for the X86 and X64 version? I know it's a dumb question but I just wanted to cover the obvious that I've overlooked assuming that some software is both x86/x64 in one installer package. Were there any hits in the system event viewer that might point to a clue? Where there any discernible error logs in the client data directory?

Some have had less issues with "unapproved" 6.2 version of the standard Boinc client. Here is a link to Berkeleys' download for the current clients (latest "recommended" is 6.2.19) in both x86 and x64 flavors.

If you're interested in another run at getting it crunching (I know it's a major pita when it locks the system instead of just crashing an app so wouldn't blame you if you're hesitant) I'd probably try:

• Get latest 6.2.19 x64 client
• Install to a folder other than "Programs" like c:\boinc
• Install as a service.

Not having had these kinds of issues, that would be my method to attempt to get it running. You might very well have tried this already and just wasted another 3 minutes reading this... If so, drinks are on me...

 

[relic]

Yeah, Nomad. I think it's worth a try. I'm new to the client but I think you may be right that it doesn't play nice with Vista x64.

I do believe I was running the 32bit 5.10.45 version. I don't recall a choice of x86 or x64 versions on download. So I'll try the "unsupported" client.

I'm not sure I like the install as a service thing, can it be monitored and paused while installed that way?

Thanks,
relic

 

[relic]

Ok...running 6.2.19 x64. It seems to pull in all the configurations from WCG.
Thanks for the advice. Lets see if this helps.

 

[nomad8u]

Yeah, Nomad. I think it's worth a try. I'm new to the client but I think you may be right that it doesn't play nice with Vista x64.

My problem is the same, being new to the client. Not to mention, I didn't have any issues getting it up and running and we both know you learn ALOT more about these things when you have to work through problems. It just stinks having to do that...

I do believe I was running the 32bit 5.10.45 version. I don't recall a choice of x86 or x64 versions on download. So I'll try the "unsupported" client.

That may very well be the issue (hopefully) and wouldn't it be nice if it were that simple. When I originally signed up and d/l'd the client this past spring, I never installed it, but from what I remember of the process, there was only one client to choose from. If you have multiple mixed OSes and IF they're looking at your system before providing a link and IF you happen to sign up on an x86 system, instant problem potential.

When I finally decided to give WCG a test drive, Aldamon had provided a link to the Boinc "generic" 6.2.18 client and suggested it was "better" than the WCG branded client so that's the way I went.

I'm not sure I like the install as a service thing, can it be monitored and paused while installed that way?

Yeah. When the client installs, and starts, it runs an app called "Boincmgr" in the systray. That your monitoring and control portal. You can "suspend" an individual core/WU or the entire project depending on what tab of the manager you're on. You can also override most of the setting that you setup online (except for what programs you're crunching but all the important settings are adjustable locally).

If you find that after getting it running (I'm only suggesting the "service" install as I've read it's less problematic) you could always uninstall and reinstall without it being as a service. At some point, all installs are going to be service installs from what I think I read somewhere. You could even grab the 5.10.45 x64 version from Berkely if you didn't want to jump right to the bleeding edge but from what little I've been able to gather, there's not many "show stopper" issues there and their Beta is already well past that latest recommended version. There's also just using the usual methods to stop the service (services.msc if that's available in Vista) like "net stop boinc"

cliffs:
grab x64 version of choice
during install deselect "use boinc screen saver"
install as a service "check protected application execution box"
install for all users "check allow all users on this computer to control boinc"
install to "generic" unprotected folder

 

[nomad8u]

Ok...running 6.2.19 x64. It seems to pull in all the configurations from WCG.
Thanks for the advice. Lets see if this helps.

Damn... you work ALOT faster than I think and type...

Glad to hear you got it sorted!

 

[relic]

The install seems pretty intuitive, I guessed the same settings you posted.
Unfortunately I had it lock up just a few minutes after install.

Maybe it's just not stable on x64 yet. I'll leave it on the other (x86) box. It runs fine there. I need this system for work and I can't have it freezing. If anyone hears of a way around the lockups let me know and I'll give it another try on Vista x64

 

[BillR]

The install seems pretty intuitive, I guessed the same settings you posted.
Unfortunately I had it lock up just a few minutes after install.

Maybe it's just not stable on x64 yet. I'll leave it on the other (x86) box. It runs fine there. I need this system for work and I can't have it freezing. If anyone hears of a way around the lockups let me know and I'll give it another try on Vista x64

Wierd, a few months ago while Stanford was sorting out it's clients I installed the BIONIC client on 3 Q6600s 2 on Vista 64 and one on XP pro. I did notice a temp rise over running the smp clients, but the BIONOC client ran fine. That was allowing BIONIC to pick it's own projects and run all out.

I'm wondering about it being a chipset issue or something equally obscure.

 

[ccityinstaller]

Wierd, a few months ago while Stanford was sorting out it's clients I installed the BIONIC client on 3 Q6600s 2 on Vista 64 and one on XP pro. I did notice a temp rise over running the smp clients, but the BIONOC client ran fine. That was allowing BIONIC to pick it's own projects and run all out.

I'm wondering about it being a chipset issue or something equally obscure.

I myself was wondering the same thing as I was reading the above posts. Relic, I don't mean to insult you, but are you running the latest chipset drivers/bios releases? I have netted over 2 million points on my o/c'd Q6600 alone without a single problem. I have been so impressed that I've installed it on a ton of family members' 'puters...Hell, one is even an old Celeron D 2.53Ghz with 512 MB ram and she chugs away nicely. I have been folding since Standford pretty much started the project (under my old username @ Anandtech before the big database crash back in 2000) and WCG is so much better organized and thought out. I think the GPU clients are wonderful but Standford doesn't seem to give a shit about releasing a stable x86 client. I still don't know why they pulled the plug on the BOINC client they were working on...

 

[relic]

The good news is that my second Vista x86 system likes the client and the XP system seems fine as well. The bad news is that the other Vista x64 freezes just like the first one did.

Oh well, I guess every client has it's issues. At least WCG works wonderfully on 32-bit Windows.