| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Beta Testing
Forum: Beta Test Support Forum
Thread: Beta_faah4000 Workunits Problem |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 27
|
|
| Author |
|
|
trongnguyen_82
Cruncher Joined: Aug 10, 2006 Post Count: 10 Status: Offline |
From lawrencehardin:
----------------------------------------The progress bar moves backwards if an 'attempt' fails. The program starts over again with a slightly different 'attempt'. From Van Fanel: As a possible way around it, I would advice to only update the progress bar when the step has been successful. In other words, instead of letting the progress bar advance in small bits, only update it after a successful iteration. The only set back is that the progress bar would seem static for most of the time... If it's the nature of the new FAAH to try several attempts until it finds a good solution to get out of the 'lengthy loop', I would prefer to see the progress bar moving back and forth instead of sticking at a number for some 'long' times. At least, it differentiate this behaviour from the bug when 'a workunit stuck at xx,xx% for several hours', in which we might have to suspend/reset BOINC to fix it.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Well I'm puzzled as I haven't seen this progress issue as yet on the few I've crunched - wondering if its dependent on boinc version?
|
||
|
|
breathesgelatin
Advanced Cruncher Joined: Aug 5, 2006 Post Count: 117 Status: Offline Project Badges: 
|
Hello trongnguyen_82, Yes, the progress indicator on these FAAH units proves in a very annoying way that we are running a non-deterministic algorithm. It is an irritating mismatch to the standard BOINC framework. I foresee lots of forum queries. I wonder why the new FAAH program seems so much more blatant about it? Is this primarily caused by the application change or by a new set of molecules? Lawrence You're probably going to have lots of people also aborting the new WUs in confusion. I really think that you need to somehow solve the problem if possible. I agree that only making it move might be a successful move - especially for the regular cruncher that might not check the forums that often. For people who crunch test WUs, seeing the progress bar move back and forth is OK once we know why, but I fear it would be too disconcerting for general users.  |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
We will take a good look at this issue before we release these versions into production. If you see my post here: https://secure.worldcommunitygrid.org/forums/wcg/viewthread?thread=19803 you will see that we have begun the process of preparing to release the new version. This process will take several days and hopefully we can get this cleared up by then.
|
||
|
|
breathesgelatin
Advanced Cruncher Joined: Aug 5, 2006 Post Count: 117 Status: Offline Project Badges:
|
We will take a good look at this issue before we release these versions into production. If you see my post here: https://secure.worldcommunitygrid.org/forums/wcg/viewthread?thread=19803 you will see that we have begun the process of preparing to release the new version. This process will take several days and hopefully we can get this cleared up by then. I think that's a good idea to review everything. I also want to note that you lose time crunched when the progress bar goes back... sometimes, not always. For example when it jumps back to 25% or 50% or wherever (depending on where it is in the process), you sometimes also lose the time crunched--sometimes as much as 45 minutes. While this doesn't really bother me in terms of helping out WCG, it will probably bother general users hoping to get full credits etc. |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges: 
|
My beta also reacts in the same way, bouncing back and fourth between 23.8% and 24.9%. The CPU time is at 15 minutes, so I guess it will continue to loop for a while. I am sure I have seen this same phenomenon several weeks ago, although I am unable to find anything related in the forum (but that only means that I have not found the right search key  ).As far as I remember it was around the same percentages, and after a few minutes it passed the 25 % mark and was completing as usual. Apparently that has not triggered a big rush of questions or problem reports... Cheers. Jean. |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
I also want to note that you lose time crunched when the progress bar goes back... sometimes, not always. For example when it jumps back to 25% or 50% or wherever (depending on where it is in the process), you sometimes also lose the time crunched--sometimes as much as 45 minutes. While this doesn't really bother me in terms of helping out WCG, it will probably bother general users hoping to get full credits etc. I need more information about this. The only reason your cpu time would go backwords is if the science app was stopped (i.e. if you shut down your client, rebooted your computer etc). In that case your cpu time would revert to the time at the last checkpoint. Can you please create a cc_config.xml file with the following options in your BOINC installation directory? <cc_config> <log_flags> <checkpoint_debug>1</checkpoint_debug> <task_debug>1</task_debug> </log_flags> </cc_config> Restart your client to start it logging. Once it has re-occurred, then please post all of the messages. Anyone experiencing this issue, please do the same. Also please make sure you are using the 5.10.45 client (you can get it here: https://secure.worldcommunitygrid.org/ms/viewDownloadAgain.do ) thanks, Kevin |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Think the percent alternation from 24.9 to 23.x and on is just the known multi docking attempts on finding best energy (graphics green en red dotted lines) as commented already by Lawrence, and the reverting indeed a client restart.
----------------------------------------The Checkpoint FAQ provides information how combined with the checkpoint log flag, post knreed, to minimize that loss, if interested to micro-manage. http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=11332 sample of the old UD agent graph which in it's BOINC graphics incarnation is much smaller: This has always been with FA@H and particular on the short units will appear amplified.... I looked at the few beta's and saw this too. But, the little green line was ever extending until near the end where it goes into the 'best energy or move on to next section' routine. [edit: Best Energy Graphic illustration refreshed showing new and old agent with explanation of red and green line]
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Apr 23, 2008 5:17:08 PM] |
||
|
|
breathesgelatin
Advanced Cruncher Joined: Aug 5, 2006 Post Count: 117 Status: Offline Project Badges:
|
I also want to note that you lose time crunched when the progress bar goes back... sometimes, not always. For example when it jumps back to 25% or 50% or wherever (depending on where it is in the process), you sometimes also lose the time crunched--sometimes as much as 45 minutes. While this doesn't really bother me in terms of helping out WCG, it will probably bother general users hoping to get full credits etc. I need more information about this. The only reason your cpu time would go backwords is if the science app was stopped (i.e. if you shut down your client, rebooted your computer etc). In that case your cpu time would revert to the time at the last checkpoint. Can you please create a cc_config.xml file with the following options in your BOINC installation directory? <cc_config> <log_flags> <checkpoint_debug>1</checkpoint_debug> <task_debug>1</task_debug> </log_flags> </cc_config> Restart your client to start it logging. Once it has re-occurred, then please post all of the messages. Anyone experiencing this issue, please do the same. Also please make sure you are using the 5.10.45 client (you can get it here: https://secure.worldcommunitygrid.org/ms/viewDownloadAgain.do ) thanks, Kevin I am using the 5.10.45 client on all my boxes. I will get you the info you need, but I don't know how to change the XML file. You're going to have to explain that more to me. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Just open it with a flat text editor like notepad. The file may already exist in either the BOINC Program or Data directory. When saving, make sure to flip the default txt extension, so it retains it's proper cc_config.xml name.
----------------------------------------You don't need to restart the client. Simply visit the Advanced menu and take the 'read config file' option to add the checkpoint message recording.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Apr 20, 2008 4:33:10 PM] |
||
|
|
|