| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Human Proteome Folding - Phase 2
Thread: HPF2 = Rosetta? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 6
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Since friday 2007-06-15 until today (that is 3 days) one of my servers was idling. I was VERY upset when I learned that I'm crunching Rosetta. I have had the same issue many times and I decided not to use rosetta@home until they debug their code. You may imagine how angry I was when I found out that rosetta is back without my knowledge.
wcg_hpf2_rosetta_5.19_i686-pc-linux-gnu -abrelax -protein lc85 -chain 7 -series 00075 -nstruct 2 Rosetta has very big issues running on Linux systems, I have experienced the same issue countless times, developers haven't been able to figure this out for a long time. What happens? A very large part of rosetta work units just hangs up, CPU is idle and other projects wait and nothing happens. One has to stop (kill) boinc client and run it again. Then, eventually the work unit finishishes sucessfuly. This is absolutely unacceptable with unnatended servers. I am very upset that I had to find out that HFP2 uses unstable rosetta engine the hard way. I want my idle CPU time to be used in some useful way, not waiting for buggy code which has nothing to do... So I suggest to state clearly that HPF2 uses Rosetta engine, so that Linux users who have bad experience with Rosetta may avoid it. I'm sorry for my quite annoyed tone, but that is just the way it is. Thank you. P.S. I joined WCG mainly for fightAIDS@home. I thought I might give other projects a go, but now it turns out that with WCG default settings you never know what your computer will eventually cruch, so I am staying strictly with fightAIDS@home. I have nothing against FPF2 project, nor Rosetta, but it is a fact, that Rosetta is not suitable for Linux systems that run permanently. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It's not a secret - HPF has always used Rosetta, and it's all explained in the Research section of the website.
When HPF2 launched, there were some major issues with the code. So WCG, being responsible and whatnot, pulled the project for as long as it took to get it fixed. And fix it they did. I haven't seen a report of a hanging Rosetta process for months and months. That's not to say it can't hang, though. But if you are having problems with it (or even just have a bad history with it) please feel no guilt about dumping it in favour of other WCG projects. It's all good! |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Hi,
----------------------------------------The highly obtuse HPF2 element was in fact the graphics and on re-launch for Linux, WCG left off the pretty picture (see link below). Runs great on windows though, very occasionally it gets stuck in a loop, but after restart, just resumes at last checkpoint and passes the hurdle on second time around without a hitch. A simple project suspend, 30 second pause and project resume is enough. No job kill necessary. When you think it's stuck check with TOP command if the science is actually running or not (presume u did, but occasionally the BOINCmgr looses sight of the BOINC daemon)). Have not heard if the Windows workaround kicks things in gear again on Linux. At any rate, it's no good on a truly unattended machine.... good news being, 3 more projects launching reasonably soon, 2 using the proven and updated Autodock engine. Resuming Human Proteome Folding 2 this weekend http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=9587 Human Proteome Folding - Phase 2 available for Linux and Mac-Intel http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=12909
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Having been inactive for a while I had hoped the 'hanging' issue was cured by now. I would love to help rosetta. But of course my first try back hung. I have now detached from it. Now i find out WCG is using it for its proteome folding proj. because it hung. I found 3 instances in my proccesses. After killing them the unit started again. I'll prob. have to skip this project.
Linux is the #2 donor for the #2 project. you'd think they'd fix this. |
|||
|
||||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Well, you have checked into Hotel California.... sit back, hook up to FA@H and GC (Genome Comparison), which goes around on my multi WCG project profiles at a 1(GC) : 8 (FA@H) ratio. Or you could just select FA@H and tick that 'alternate work' box at bottom of Device Profile in case FA@H has a distribution problem or simply does not have work for a platform (very temporarily). Occasionally people report this event, but if you set the task buffer to e.g. 2 days it should in principle not happen.
----------------------------------------edit: my space bar fails regularly.... need some keyboard oil.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Jun 20, 2007 9:07:59 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello osbie,
Just about anything can go wrong at rare intervals, but you are reporting a recurrent problem that occurs frequently using multiple versions of Rosetta. Yet, restarting can cause the formerly frozen work unit to finish successfully. If I were you, I would avoid running Rosetta. Better safe than sorry. Thinking about causes - - BOINC will freeze a work unit if it exceeds the memory space allowed during execution. Rosetta on HPF2 is potentially much larger than FAAH or GC (our smallest project). The size depends on the work unit. The question is whether BOINC pays any attention to other programs running on the computer. I wouldn't think that it would, so that would not account for finishing a frozen work unit after a restart.  Oh well. . . Thanks for all your crunching on FAAH! Lawrence |
||
|
|
|