| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Human Proteome Folding
Thread: Errors with EP jobs and BOINC Client on Windows |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 10
|
|
| Author |
|
|
clehardy
Cruncher Joined: Apr 18, 2006 Post Count: 6 Status: Offline |
Every time one of my computers submits a result set for an EP job (ep259_02 for instance), the result is an error. However, if that same computer submits a result set for FAAH (faah0451_ d016cb175_ x1hpv_ 00 for instance), it's valid.
OTHER computers are able to do both projects just fine. It's just this one computer that can't seem to get them processed fine. So... what might be the problem? The computer is just processing jobs, not doing anything else (pretty much anyway) and has no viruses, spyware, anything else on it (I check it fairly frequently) and it's a server-class machine with a HT processor (and ECC memory of course). So it's also a stable machine... So, what's going on? Thanks for the help, Charlie |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello clehardy,
This does not sound like anything that I have run across before. But, reasoning from first principles, here is something to try. BOINC keeps a copy of Rosetta and AutoDock in the Projects subdirectory that it copies into the working Slot subdirectory. Maybe something corrupted Rosetta. So try detaching from WCG, then re-attaching in order to download new copies of the project applications. It's the only thing that occurs to me. Lawrence |
|||
|
||||
|
clehardy
Cruncher Joined: Apr 18, 2006 Post Count: 6 Status: Offline |
Thanks for your response... I also got a response from one of the World Community Grid people saying that it looks like my computer has some other process that's taking up the CPU and is causing BOINC to shut the job down abnormally.
Apparently it's happening for both of them (FAAH and EP) however, FAAH is handling the error and EP isn't. I do have a VMware machine running on that same box that's doing processing with BOINC itself so I shut that down and we'll see if that fixes the problem :-) If not, I'll try your solution next. Thanks! Charlie |
||
|
|
David Autumns
Ace Cruncher UK Joined: Nov 16, 2004 Post Count: 11062 Status: Offline Project Badges: 
|
Hi clehardy
----------------------------------------I've just been talking with Lawrence and I have the same issue. I have just 224Mb of Ram after 32Mb has been taken for graphics duties and I'm running Win 2K3 Web Edition on this box. Like you I just have antivirus and antispyware running. I was wondering if your machine was similarly spec'ed? It's only this box thaty shows these symptoms I've just taken a hoover to the CPU heatsink it was a tad dusty but I have to confess I've seen worse. I'll report back on my findings Dave  |
||
|
|
clehardy
Cruncher Joined: Apr 18, 2006 Post Count: 6 Status: Offline |
No... my box is a bit more beefy (2.8 Ghz P4 (hyper-threaded) with 2 GB of DDR2 ECC RAM). So that's not my problem...
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hmmm....
I don't know how much information you've already given to the WCG techs, but they have access to more data than we do here on the forum. So, you'll have to tell us more: are the work units failing part way through, or are they completing normally? Check the error log, and if possible copy it here. From what the tech told you, it sounds like the result is not completing normally. There are a few possibilities here, but to diagnose a conflict, we need as much information about the computer as possible. What other software does it run? Antivirus and security software are common culprits. Are you sure it's not a DEP problem? I believe your CPU supports hardware DEP, and that may need an exception adding for BOINC. What are your current DEP settings? |
||
|
|
clehardy
Cruncher Joined: Apr 18, 2006 Post Count: 6 Status: Offline |
If you're talking to me.... My problem is fixed now. Everything works great for me now that I closed down the BOINC process running on a VMware machine I had (which is what the techs recommended). If you're talking to Autumns, then thanks for your help and have fun! :-)
|
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
clehardy,
----------------------------------------Ah - that explains what was going on. I discovered this the other day (I use vmware to do all of my Linux development work). If you run BOINC inside an operating system running on top of VMWARE then you have to remember that within that guest operating system BOINC will be run at lowest priority and properly swap with other applications in that environment. However the entire VMWARE environment is a single process running at normal priority in the host operating system. This means that the VMWARE process will be asking for as much runtime as the host operating system will give it and since it is running at a normal priority it will interfere with other applications set to the same priority. This means that if you have a BOINC instance inside the vmware environment and a BOINC instance running on the host operating system, then the BOINC instance running on the host operating system will get almost no execution time becuase it is a lower priority then the vmware process. I hope this makes sense - let me know if you or anyone else has futher questions. Kevin [Edit 1 times, last edit by knreed at May 9, 2006 1:40:07 AM] |
||
|
|
clehardy
Cruncher Joined: Apr 18, 2006 Post Count: 6 Status: Offline |
Right, that does make sense and is what was happening. I of course also can't lower the priority of the VMware instance so that it runs below BOINC on the host machine (or raise BOINC on the host machine) because then either the guest VMware machine would starve of processing power OR... the whole machine would be starved of processing power because BOINC on the host is too high.
Either way, it's a lose-lose situation. So, I chose to stop BOINC on the guest and leave it running on the host (since the guest version of BOINC wasn't really doing all that much anyway). Thanks again for your help, Charlie |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Well, clehardy: running it only on the host is the best solution, since it will get all the cycles not used by anything else. There was no possible benefit you could have got by running it on the client too.
So it's win-win :-) |
||
|
|
|